Planet SciPy

This Week in Machine Learning: Oceans, Marketers, and Wildlife Protection

Machine learning is fascinating. New things happen every second while we’re busy performing our daily tasks. If you want to know what big things have happened over the last week, make sure to check this weekly roundup! Here are the best picks from the last week from the world of machine learning. Enjoy the read! […]

The post This Week in Machine Learning: Oceans, Marketers, and Wildlife Protection appeared first on neptune.ai.

How to Do Data Exploration for Image Segmentation and Object Detection (Things I Had to Learn the Hard Way)

I’ve been working with object detection and image segmentation problems for many years. An important realization I made is that people don’t put the same amount of effort and emphasis on data exploration and results analysis as they would normally in any other non-image machine learning project.   Why is it so?  I  believe there are two […]

The post How to Do Data Exploration for Image Segmentation and Object Detection (Things I Had to Learn the Hard Way) appeared first on neptune.ai.

Technical discussions are hard; a few tips

A large software project is above all a social exercise in which technical experts try to reach good decisions together, for instance on github pull requests. But communication is difficult, in …

This Week in Machine Learning: Bees, Sky Objects, & HMI

Every day brings new opportunities. A week brings even more. To make sure you’re getting the most out of it, we’ve gathered a list of trending articles of the week – everything about Data Science, AI, tech, and machine learning. Here are the best picks from the last week from the world of the machine […]

The post This Week in Machine Learning: Bees, Sky Objects, & HMI appeared first on neptune.ai.

The Best Comet.ml Alternatives

Comet is one of the most popular tools used by people working on machine learning experiments. It is a self-hosted and cloud-based meta machine learning platform allowing data scientists and teams to track, compare, explain, and optimize experiments and models. Comet proposes an open-source Python library to allow data scientists to integrate their code with […]

The post The Best Comet.ml Alternatives appeared first on neptune.ai.

The Best Tools for Machine Learning Model Visualization

The phrase “Every model is wrong but some are useful” is especially true in Machine Learning. When developing machine learning models you should always understand where it works as expected and where it fails miserably. There are many methods that you can use to get that understanding: Look at evaluation metrics (also you should know […]

The post The Best Tools for Machine Learning Model Visualization appeared first on neptune.ai.

Random Forest Regression: When Does It Fail and Why?

In this article, we’ll look at a major problem with using Random Forest for Regression which is extrapolation.  We’ll cover the following items: Random Forest Regression vs Linear Regression Random Forest Regression Extrapolation Problem Potential solutions Should you use Random Forest for Regression? Let’s dive in.  Random Forest Regression vs Linear Regression Random Forest Regression is […]

The post Random Forest Regression: When Does It Fail and Why? appeared first on neptune.ai.

The Best Tools to Visualize Metrics and Hyperparameters of Machine Learning Experiments

Evaluating your model on the key metrics is a crucial first step in understanding your model quality. Keeping track of hyperparameters and corresponding evaluation metrics is important because small changes in hyperparameters can sometimes have a big impact on model quality. And so, understanding which hyperparameters have an impact and which do not affect evaluation […]

The post The Best Tools to Visualize Metrics and Hyperparameters of Machine Learning Experiments appeared first on neptune.ai.

Tidynamics, what use?

In 2018 I published small Python library, tidynamics. The scope was deliberately limited: compute the typical correlation functions for stochastic and molecular dynamics: the autocorrelation and the mean-square displacement. Two years later, I wonder about its usage.

Moderna, IMC Renew NumFOCUS Corporate Sponsorships

Monday, May 18th, 2020 Two NumFOCUS corporate supporters recently made fresh commitments to our open source mission. Trading firm IMC and biotechnology company Moderna Therapeutics each renewed their corporate sponsorships earlier this month. Both companies have supported NumFOCUS since 2018 at our Silver and Bronze sponsorship levels, respectively. Asked about his company’s decision to partner […]

The post Moderna, IMC Renew NumFOCUS Corporate Sponsorships appeared first on NumFOCUS.

This Week in Machine Learning: Amazon Releases Kendra, Brain-Inspired Algorithms, Confused AI Models & More

Machine learning is fascinating. New things happen every second while we’re busy performing our daily tasks. If you want to know what big things have happened over the last week, make sure to check this weekly roundup! Here are the best picks from the last week from the world of the machine learning. Enjoy the […]

The post This Week in Machine Learning: Amazon Releases Kendra, Brain-Inspired Algorithms, Confused AI Models & More appeared first on neptune.ai.

NumFOCUS Projects helping combat the COVID-19 pandemic

Open source tools are uniquely positioned to help combat the ongoing COVID-19 pandemic through their adaptable and collaborative nature. NumFOCUS sponsored and affiliated projects are being used on a global scale to meet the needs of researchers and data scientists. Our projects are being used in groundbreaking scientific efforts to create response models, visualize and […]

The post NumFOCUS Projects helping combat the COVID-19 pandemic appeared first on NumFOCUS.

Top Open Source Tools and Libraries for Deep Learning – ICLR 2020 Experience

Where is cutting-edge deep learning created and discussed? One of the top places is ICLR – a leading deep learning conference, that took place on April 27-30, 2020. As a fully virtual event, with 5600+ participants and almost 700 papers/posters it could be called a great success. You can find comprehensive info about the conference […]

The post Top Open Source Tools and Libraries for Deep Learning – ICLR 2020 Experience appeared first on neptune.ai.

Lazy River of Curious Content 0

This is the first post of what I'm calling a Lazy River of Curious Content. This is a way to review stuff that I've been doing, dealing with, or find interesting during the week recently (This was originally written two weeks ago, May 3rd, my shoddy internet connectivity kept me from posting it.). I'm loosely following the format that Justin Sherrill uses with great effect over at https://dragonflydigest.com

Learn NixOS by turning a Raspberry Pi into a Wireless Router Friend of the show, Anthony Scopatz, tried NixOS for the first time and provides a detailed report:

"While I had read the NixOS pamphlets, and listened politely when the faithful came knocking on my door at inconvenient times, I had never walked the path of functional Linux enlightenment myself"

Reading through that made me file away a todo of writing up how I use propellor (and why). But those todo sometimes just pile up for a while...

An interview of one of my long time nerd-crushes, Rob Pike. The questions focus on the Go programming

Interview with a Chief AI Scientist: Arash Azhand

Some time ago I had a chance to interview a great artificial intelligence researcher and Chief AI Scientist in Lindera, Arash Azhand. We talked about: The AI technology behind his work at Lindera His career path How it is to be a research-centered scientist How to become a good leader Why it is important to […]

The post Interview with a Chief AI Scientist: Arash Azhand appeared first on neptune.ai.

sourmash databases as zip files, in sourmash v3.3.0

The feature that I'm most excited about in sourmash 3.3.0 is the ability to directly use compressed SBT search databases.

Previously, if you wanted to search (say) 100,000 genomes from GenBank, you'd have to download a several GB .tar.gz file, and then uncompress it out to ~20 GB before searching it. The time and disk space requirements for this were major barriers for teaching and use.

In v3.3.0, Luiz Irber fixed this by, first, releasing the niffler Rust library with Pierre Marijon, to read and write compressed files; second, replacing our old khmer Bloom filter nodegraph with a Rust implementation (sourmash PR #799); and, third, adding direct zip file storage (sourmash #648).

So, as of the latest release, you can do the following:

# install sourmash v3.3.0
conda create -y -n sourmash-demo \
    -c conda-forge -c bioconda sourmash=3.3.0

# activate environment
conda activate sourmash-demo

# download the 25k GTDB release89 guide database (~1.4 GB)
curl -L https://osf.io/5mb9k/download > gtdb-release89-k31.sbt.zip

# grab

LaKademy 2019

Akademy 2019

Highlights of the Ibis 1.3 release

Yellowbrick Update – April 2020

Yellowbrick released Version 1.1 on February 25, 2020.  If you haven’t yet upgraded simply type pip install yellowbrick -U or conda install -c districtdatalabs yellow-brick into your terminal/command prompt to get it.  The major improvement in v1.1 is introducing quick methods or one-liners to generate your favorite ML plots more quickly with Yellowbrick.  Dr.  Rebecca […]

The post Yellowbrick Update – April 2020 appeared first on NumFOCUS.

2020 PyData Conferences Update [COVID-19]

We wanted to give an update to our community regarding the upcoming 2020 PyData conferences. We have been closely monitoring the situation and to help ensure the safety of our community given the threat of the COVID-19 virus, the following in-person events have been postponed to 2021: PyData Miami PyData Amsterdam PyData LA PyData London PyData […]

The post 2020 PyData Conferences Update [COVID-19] appeared first on NumFOCUS.

Scientific Software Developer- Contract Basis [SunPy Project]

Scientific Software Developer- Contract Basis NumFOCUS is seeking a Scientific Software Developer to support the SunPy project. SunPy is a Python-based open source scientific software package supporting solar physics data analysis. This is a 1 year contract.     The successful applicant will work to improve SunPy’s functionality. There are four main tasks:   Report on […]

The post Scientific Software Developer- Contract Basis [SunPy Project] appeared first on NumFOCUS.

Thanking the people behind Spyder 4

Creating the ultimate terminal experience in Spyder 4 with Spyder-Terminal

This blogpost was originally published on the Quansight Labs website.

The Spyder-Terminal project is revitalized! The new 0.3.0 version adds numerous features that improve the user experience, and enhances compatibility with the latest Spyder 4 release, in part thanks to the improvements made in the xterm.js project.

First, we were able to update all the old JavaScript files to use ES6/JSX syntax and the tests for the client terminal. This change simplified the code base and maintenance and allows us to easily extend the project to new functionalities that the xterm.js API offers. In order to compile this code and run it inside Spyder, we migrated our deployment to Webpack.

In the new release, you now have the ability to configure which shell to use in the terminal. On Linux and UNIX systems, bash, sh, ksh, zsh, csh, pwsh, tcsh, screen, tmux, dash and rbash are supported, while cmd and powershell are the

Introducing ndindex, a Python library for manipulating indices of ndarrays

Software and workflow development practices (April 2020 update)

Over the last 10-15 years, I've blogged periodically about how my lab develops research software and build scientific workflows. The last update talked a bit about how we've transitioned to snakemake and conda for automation, but I was spurred by an e-mail conversation into another update - because, y'all, it's going pretty well and I'm pretty happy!

Below, I talk through our current practice of building workflows and software. These procedures work pretty well for our (fairly small) lab of people who mostly work part-time on workflow and software development. By far the majority of our effort is usually spent trying to understand the results of our workflows; except in rare cases, I try to guide people to spend at most 20% of their time writing new analysis code - preferably less.

Nothing about these processes ensures that the scientific output is correct or useful, of course. While scientific correctness of computational workflows necessarily depends (often critically) on the correctness of the

LaKademy 2019

PyTorch TensorIterator Internals

Is it getting better yet? An optimistic visual guide to the Coronavirus pandemic

As the apocalypse rumbles on, I found myself wondering "Is it getting any better?"

Daily updates of spiralling case numbers (and worse, deaths) does little to give a sense of whether we're getting to, or already past, the worst of it.

To answer that question for myself and you, I …

How to give a bad online talk

Today at lab meeting, I wanted to brainstorm about how to give good online talks, because I'm giving a few remote talks in the next month. Tracy suggested that perhaps I should demonstrate a bad talk first, just to get everyone on the same page.

So I did!

Direct (YouTube link)

...enjoy? It's short, and not TOO painful if you show up with low expectations!

First, let me say that we were tremendously ...inspired by Greg Wilson's How to Teach Badly and How to Teach Badly (part 2)!

So here's what I did --

I put together a few slides on some stuff that I'd been working on recently, so it would look reasonable.

My initial screen opened with a private Twitter message up, to mimic inadvertent content sharing :).

I started out with "I didn't have a lot of time to prepare for this meeting so apologies for some of the slides."

My slide theme was very hard to read - bad fonts and colors.

A

On the Link Between Polynomials and Optimization

There's a fascinating link between minimization of quadratic functions and polynomials. A link that goes deep and allows to phrase optimization problems in the language of polynomials and vice versa. Using this connection, we can tap into centuries of research in the theory of polynomials and shed new light on …

pheriday 3: infrastructure

Documentation as a way to build Community

PyData COVID-19 Response

The safety and well-being of our community are extremely important to us. We have therefore decided to postpone all PyData conferences scheduled to take place until the end of June: PyData Miami PyData London PyData Amsterdam We have been closely monitoring the situation and believe this is the best action to take based on the […]

The post PyData COVID-19 Response appeared first on NumFOCUS.

uarray: GSoC Participation

Planned architectural work for PyData/Sparse

Statement on Coronavirus

As you are aware, the Coronavirus (COVID-19) is a topic of frequent and ongoing discussions. We would like to provide an update on our status and policies as well as provide resources for additional information. As of today, our event schedule remains as posted on event sites. Any changes or updates will be immediately shared. […]

The post Statement on Coronavirus appeared first on NumFOCUS.

Some snakemake hacks for dealing with large collections of files

This winter quarter I taught my usual graduate-level introductory bioinformatics lab at UC Davis, GGG 201(b), for the fourth time. The course lectures are given by Megan Dennis and Fereydoun Hormozdiari, and I do a largely separate lab that aims to teach the basics of practical variant calling, de novo assembly, and RNAseq differential expression.

I also co-developed and co-taught a new course, GGG 298 / Tools for Data Intensive Research, with Shannon Joslin, a graduate student here in Genetics & Genomics who (among other things) took GGG 201(b) the first time I offered it. GGG 298 is a series of ten half-day workshops where we teach shell, conda, snakemake, git, RMarkdown, etc - you can see the syllabus for GGG 298 here.

This time around, I did a complete redesign of the GGG 201(b) lab (see syllabus) to focus on using snakemake workflows.

I'm 80% happy with how it went - there's some overall fine tuning to be done, and snakemake has some corners that need more explaining than other corners, but I think the

Akademy 2019

My Unexpected Dive into Open-Source Python

Announcing JupyterCon 2020

NumFOCUS is excited to be a part of JupyterCon 2020. JupyterCon will be held August 10 – 14 in Berlin, Germany at the Berlin Conference Center. We invite you to participate in this exciting community event! Read the full announcement here. JupyterCon 2020 is an event brought to you in partnership by Project Jupyter and NumFOCUS.

The post Announcing JupyterCon 2020 appeared first on NumFOCUS.

MDAnalysis joins NumFOCUS Sponsored Projects

NumFOCUS is pleased to announce the newest addition to our fiscally sponsored projects: MDAnalysis MDAnalysis is a Python library for the analysis of computer simulations of many-body systems at the molecular scale, spanning use cases from interactions of drugs with proteins to novel materials. It is widely used in the scientific community and is written […]

The post MDAnalysis joins NumFOCUS Sponsored Projects appeared first on NumFOCUS.

Two talks at JGI in May: sourmash, spacegraphcats, and disease associations in the human microbiome.

Hello all! I'm giving two metagenomics talks - a tech talk and a bio talk - at the Joint Genome Institute on May 7, 2020. The abstracts are below.

The JGI just moved to a new building at LBNL, so these talks are much more accessible to the UC Berkeley and LBNL communities than they would have been a year ago. I hope interested people can make it!

The talks will be in the afternoon on May 7th at the Integrative Genomics Building, LBNL Bldg 91-310. I've put the tentative times down. I'll update this post with final times and contact information for security + parking passes closer to the day.

(This will be the Science and Technology seminar, 3-4pm on May 7.)

Inflammatory bowel disease (IBD) is a spectrum of diseases characterized by chronic inflammation of the intestines; it is likely caused by host-mediated inflammatory responses at least in part elicited by microorganisms. As of 2015, 1.3% of US adults have been diagnosed

Creating the ultimate terminal experience in Spyder 4 with Spyder-Terminal

Advancing research software in the UK through an SSI fellowship

I have selected as part of the 2020 cohort of Fellows of the Software Sustainability Institute!

The Institute cultivates world-class research with software. It's based at the universities of Edinburgh, Manchester, Southampton, and Oxford in the UK. Their motto says it all:

The SSI has a yearly fellowship program to fund the organization of communities around scientific software (creating of local user groups, workshops, hackathons, etc). Even more importantly, they organize several events to get current and past fellows in the same place doing awesome stuff. I'm really looking forward to this year's Collaborations Workshop (registration is open to all, not just fellows). I applied at the end of last year and was selected to join the 2020 cohort of fellows along with some truly amazing people.

My plan for the fellowship is to run

Don't fund Software that doesn't exist

sourmash-oddify: a workflow for exploring contamination in metagenome-assembled genomes

(Thanks to Erich Schwarz, Taylor Reiter, and Donovan Parks for brainstorming and feedback on this stuff. Thanks also to Luiz Irber and Phillip Brooks for their work on sourmash!)

Yesterday, I posted about using k-mers and taxonomy to investigate Genbank genomes for potential contamination.

The underlying idea is pretty simple: look for subsets of k-mers that don't match the inferred taxonomy of the genome bins they're from, then analyze.

What started me down this path over two years ago (!!) was the use of the same underlying Tara Oceans metagenomic data for two separate papers, Tully et al., 2018 and Delmont et al., 2018. Both groups released their data early along with bioRxiv preprints, and it proved to be a treasure trove for my bioinformatics methods development - all of the sourmash lca functionality as well a lot of other functionality came from a series of about 14 blog posts examining these genomes.

I last left off with the

Finding problematic bacterial/archaeal genomes using k-mers and taxonomy

(Happy New Year, everyone! Thanks on this blog post go out to Erich Schwarz and Taylor Reiter, for offering helpful suggestions and asking tough questions as I meandered through this work!)

Yesterday, I posted about using sourmash lca classify to taxonomically classify bacterial and archaeal genomes quickly, and compared the results to the full GTDB taxonomy. The tl;dr was that sourmash works pretty well and returns results consistent with GTDB and GTDB-Tk, but that it often doesn't classify as precisely as GTDB-Tk.

I was kind of expecting that at the species level, because there is a limit to the kind of precision that downsampled k-mers can achieve: the last 1-0.1% of nucleotide similarity can be a bit wobbly with sourmash (<- technical term).

But I was surprised to see the phylum and superkingdom level limits. sourmash lca classify couldn't classify 235 genomes beyond phylum level! What could be causing this?

I took a closer look at GCF_001477405, a genome tagged as Staphylococcus sciuri in Genbank. Using sourmash lca summarize,

How does sourmash's lca classification routine compare with GTDB classifications?

Yesterday I posted about the GTDB taxonomy; we are now providing prepared databases that can be used with sourmash's taxonomy classification routines to classify genomes with GTDB.

The databases we posted are built from the dereplicated 25k GTDB genomes distributed as part of the GTDB-Tk classification toolkit, and not the full 145k classifications in GTDB. So they are smaller than they could be, and also potentially lower resolution. Moreover, sourmash uses k-mers instead of amino acids, which may lead to different classifications.

A good first question is, how well do classifications with sourmash lca classify & 25k genomes compare to the full 145k classifications in GTDB? This is basically a measure of generalizability - how reliably can we infer the classifications of the 145k genomes from the 25k?

I classified all 420k Genbank genomes using sourmash lca classify with k=31, and I then wrote a script to compare the output to the GTDB taxonomy. This involved some rather nasty identifier conversion which sometimes failed, but we ended up with a good

Sourmash LCA databases now available for the GTDB taxonomy

I am happy to announce that we have made available prepared sourmash taxonomy ("LCA") databases for release 89 of the GTDB taxonomy.

The databases are available for download from the Open Science Framework in this project. There are prepared databases avaialble for k=21, k=31, and k=51.

GTDB is a revised bacterial and archaeal taxonomy based on phylogenetic relations between proteins from approximately 25k genomes. You can read more about it here.

GTDB is an alternative to the NCBI taxonomy. It is used by (among others) MGnify, the EBI metagenomics resource.

Sourmash is a research platform and bioinformatics tool for searching and analyzing genomes, based on a MinHash-inspired approach that allows genome similarity searches, genome containment searches, and compositional analysis of k-mers in large sequence data sets. You can read more about it here.

There are three immediate uses for these databases:

• you can use the sourmash lca classify routine (and other LCA commands) to do taxonomic classification of genomes

metadsl PyData talk

class Number(metadsl.Expression):
   @metadsl.expression
   def __add__(self, other: Number) -> Number:
       ...

   @metadsl.expression
   @classmethod
   def from_int(cls, i: int) -> Number:
       ...


@metadsl.rule
def add_zero(y: Number):
   yield Number.from_int(0) + y, y
   yield y + Number.from_int(0), y

Two PhD studentships at the University of Liverpool

I have two open positions for funded studentships at the University of Liverpool. Applications are open until 10 January 2020.

Follow the links for more detailed versions.

Bringing machine learning techniques to geophysical data processing

The goal of this project is to investigate the use of existing machine learning techniques to process gravity and magnetics data using the Equivalent Layer Method. The methods and software developed during this project can be applied to process large amounts of gravity and magnetics data, including airborne and satellite surveys, and produce data products that can enable further scientific investigations. Examples of such data products include global gravity gradient grids from GOCE satellite measurements, regional magnetic grids for the UK, gravity grids for the Moon and Mars, etc.

Large-scale mapping of the thickness of the

SunPy Receives NASA Grant, Helps Generate Parker Solar Probe Results

The post SunPy Receives NASA Grant, Helps Generate Parker Solar Probe Results appeared first on NumFOCUS.

NumFOCUS Kicks Off Year-End Fundraising with $2,500 Gift

Giving Tuesday 2019 marked the start of NumFOCUS’s year-end fundraising campaign, and this year’s effort began with a major donation from the organization’s Board President, Andy Terrel. “Today I’m pledging a gift of $2,500 to help kick off the NumFOCUS end-of-year fundraising campaign,” Terrel wrote yesterday in an e-mail to the NumFOCUS community. “I believe […]

The post NumFOCUS Kicks Off Year-End Fundraising with $2,500 Gift appeared first on NumFOCUS.

Stepping Onto a New Path…

The post Stepping Onto a New Path… appeared first on NumFOCUS.

Getting a big scientific prize for open-source software

A few days ago, Loïc Estève, Alexandre Gramfort, Olivier Grisel, Bertrand Thirion, and myself received the “Académie des Sciences Inria prize for transfer”, for our contributions to the scikit-learn project …

Variable Explorer improvements in Spyder 4

This blogpost was originally published on the Quansight Labs website.

Spyder 4 will be released very soon with lots of interesting new features that you'll want to check out, reflecting years of effort by the team to improve the user experience. In this post, we will be talking about the improvements made to the Variable Explorer.

These include the brand new Object Explorer for inspecting arbitrary Python variables, full support for MultiIndex dataframes with multiple dimensions, and the ability to filter and search for variables by name and type, and much more.

It is important to mention that several of the above improvements were made possible through integrating the work of two other projects. Code from gtabview was used to implement the multi-dimensional Pandas indexes, while objbrowser was the foundation of the new Object Explorer.

For Spyder 4 we added a long-requested feature: full support for inspecting any kind of Python object through the Variable

File management improvements in Spyder 4

This blogpost was originally published on the Quansight Labs website.

Version 4.0 of Spyder is almost ready! It has been in the making for well over two years, and it contains lots of interesting new features. We will focus on the Files pane in this post, where we've made several improvements to the interface and file management tools.

In order to simplify the Files pane's interface, the columns corresponding to size and kind are hidden by default. To change which columns are shown, use the top-right pane menu or right-click the header directly.

First, we added the ability to associate different external applications with specific file extensions they can open. Under the File associations tab of the Files preferences pane, you can add file types and set the external program used to open each of them by default.

Once you've set this up, files will automatically launch in the associated application when opened from the Files pane in Spyder.

This Week in Glean: November 1, 2019

When data goes wrong

Loan Amortisation Schedule using R and Python

Introducing napari: a fast n-dimensional image viewer in Python

O SERPRO e a validação de documentos digitais

Os documentos digitais (no plural) do Brasil

How to Evaluate the Logistic Loss and not NaN trying

A naive implementation of the logistic regression loss can results in numerical indeterminacy even for moderate values. This post takes a closer look into the source of these instabilities and discusses more robust Python implementations.

Uvas Gold 200

My poem about a rainy 200k was published in the Fall 2019 issue of American Randonneur (a quarterly magazine published by Randonneurs USA)

I've been doing samizdat poetry for as long as I've had a web presence (since 1999), but I am now officially a published poet! (I am deliberately not counting the embarrasing hackjob that was published in a youth anthology when I was in 8th grade.)

You can find "Uvas Gold 200" on page 26 - either directly on this skeuomorphic leafing viewer or the PDF, but I'm republishing both the exposition blurb and the poem below. If you prefer to listen, I recorded a reading of it that you can download in different flavors: a local audio only, a local video, or the embeded video version below.

Uvas Gold 200k starts and ends in Fremont, CA and was held on Saturday, December 1st, 2018. The ride frontloads the climbing by going nearly half-way up Mount

Grupo de Estudos do Laboratório Amazônico de Estudos Sociotécnicos – UFPA

pelas ruas de Belém

Spyder 4.0: Kite integration is here

This blogpost was originally published on the Quansight Labs website.

Note: Kite is sponsoring the work discussed in this blog post, and in addition supports Spyder 4.0 development through a Quansight Labs Community Work Order.

As part of our next release, we are proud to announce an additional completion client for Spyder, Kite. Kite is a novel completion client that uses Machine Learning techniques to find and predict the best autocompletion for a given text. Additionally, it collects improved documentation for compiled packages, e.g. Matplotlib, NumPy and SciPy, that cannot be obtained easily by using traditional code analysis packages such as Jedi. Although Kite is not open source like Spyder, you can download it without charge at the Kite website.

By incorporating Kite into Spyder, we will improve and provide the ultimate autocompletion and signature retrieval experience for most of the scientific Python stack and beyond. For instance, let’s take a look at the following PyTorch completion. While

An initial report on the Common Fund Data Ecosystem

For the past 6 months or so, I've been working with a team of people on a project called the Common Fund Data Ecosystem. This is a targeted effort within the NIH Common Fund (CF) to improve the Findability, Accessibility, Interoperability, and Reusability - a.k.a. "FAIRness" - of the data sets hosted by their Data Coordinating Centers.

(You can see Dr. Vivien Bonazzi's presentation if you're interested in more details on the background motivation of this project.)

I'm thrilled to announce that our first report is now available! This is the product of a tremendous data gathering effort (by many people), four interviews, and an ensuing distillation and writing effort with Owen White and Amanda Charbonneau. To quote,

This assessment was generated from a combination of systematic review of online materials, in-person site visits to the Genotype Tissue Expression (GTEx) DCC and Kids First, and online interviews with Library of Integrated Network-Based Cellular Signatures (LINCS) and Human Microbiome Project

Object Oriented Programming in Python : Learn by Examples

Precision Recall Curve Simplified

Comparing two genome binnings quickly with sourmash

tl;dr? Compare and cluster two collections of 1000+ metagenome-assembled genomes in a few minutes with sourmash!

A week ago, someone e-mailed me with an interesting question: how can we compare two collections of genome bins with sourmash?

Why would you want to do this? Well, there's lots of reasons! The main one that caught my attention is comparing genomes extracted from metagenomes via two different binning procedures - that's where I started almost two years ago, with two sets of bins extracted from the Tara ocean data. You might also want to merge bins that were similar to produce a (hopefully) more complete bin, or you could intersect bins that were similar to produce a consensus bin that might be higher quality, or you could identify bins that were in one collection and not in the other, to round out your collection.

I'm assuming this is done by lots of workflows - I note, for example, that the metaWRAP workflow includes a 'bin refinement' step that must do something like

Calculate KS Statistic with Python

A Complete Guide to Python DateTime Functions

In this tutorial, we will cover python datetime module and how it is used to handle date, time and datetime formatted columns (variables). It includes various practical examples which would help you to gain confidence in dealing dates and times with python functions. In general, Date types columns are not easy to manipulate as it comes with a lot of challenges like dealing with leap years, different number of days in a month, different date and time formats or if date values are stored in string (character) format etc.

Table of Contents

---

Introduction : datetime moduleIt is a python module which provides several functions for dealing with dates and time. It has four classes as follows which are explained in the latter part of this article how these classes work.

1. datetime
2. date
3. time
4. timedelta

People who have no experience of working with real-world datasets might have not encountered date columns. They might be under impression that working with dates is rarely used and not so

What are *args and **kwargs and How to use them

Python : 10 Ways to Filter Pandas DataFrame

Python Dictionary Comprehension with Examples

d = {'a': [1,2], 'b': [3,4], 'c': [5,6]}

d.keys() # 'a', 'b', 'c'
d.values() # [1, 2], [3, 4], [5,

HTML outputs in Jupyter

User interaction in data science projects can be improved by adding a small amount of visual deisgn.

To motivate effort around visual design we show several simple-yet-useful examples. The code behind these examples is small and accessible to most Python developers, even if they don’t have much HTML experience.

This post in particular focuses on Jupyter’s ability to add HTML output to any object. This can either be full-fledged interactive widgets, or just rich static outputs like tables or diagrams. We hope that by showing examples here we will inspire some throughts in other projects.

This post was supported by replies to this tweet. The rest of this post is just examples.

I originally decided to write this post after reading another blogpost from the UK Met office, where they included the HTML output of their library Iris in a a blogpost

(work by Peter Killick, post by Theo McCaie)

The fact that the output provided by an interactive session is the same output that you would provide in a published result helps everyone. The interactive

Python list comprehension with Examples

Don't cite the No Free Lunch Theorem

15 ways to read CSV file with pandas

import pandas as pd

Matplotlib Tutorial : Learn with Examples in 3 hours

This tutorial outlines how to perform plotting and data visualization in python using Matplotlib library. The objective of this post is to get you familiar with the basics and advanced plotting functions of the library. It contains several examples which will give you hands-on experience in generating plots in python.

Table of Contents

---

What is Matplotlib?It is a powerful python library for creating graphics or charts. It takes care of all of your basic and advanced plotting requirements in Python. It took inspiration from MATLAB programming language and provides a similar MATLAB like interface for graphics. The beauty of this library is that it integrates well with pandas package which is used for data manipulation. With the combination of these two libraries, you can easily perform data wrangling along with visualization and get valuable insights out of data. Like ggplot2 library in R, matplotlib library is the grammar of graphics in Python and most used library for charts in Python.

Basics

Write Short Blogposts

I encourage my colleagues to write blogposts more frequently. This is for a few reasons:

1. It informs your broader community what you’re up to, and allows that community to communicate back to you quickly.

   You communicating to the community fosters a sense of collaboration, openness, and trust. You gain collaborators, build momentum behind your work, and curate a body of knowledge that early adopters can consume to become experts quickly.

   Getting feedback from your community helps you to course-correct early in your work, and stops you from wasting time in inefficient courses of action.

   You can only work for a long time without communicating if you are either entirely confident in what you’re doing, or reckless, or both.

2. It increases your visibility, and so is good for your career.

   I have a great job. I find my work to be both

How to encourage participation in teleconferences

(and/or how to run effective teleconferences!)

I participate in a lot of teleconferences, and some of them aren't very participatory, for various reasons. Recently a good friend asked for suggestions on how to open up the phone calls, and I came up with the below ideas. What am I missing? What did I get wrong?

First, post a meeting agenda with a medium amount of detail, well in advance ( > 24 hours).

• Posting an agenda in advance gives people time to think about things, if they are interested.
• The medium amount of detail (up to a paragraph) lets people understand what it’s about, see what the major issues/questions are, and think of questions or comments they may have.
• If the agenda is posted > 24 hours in advance, you can reasonably expect people to have read it, and if people want to add things to the agenda on the call you punt them to the next call instead.

Basically, if you spring a skeleton

How to drop one or multiple columns from Pandas Dataframe

In this tutorial, we will cover how to drop or remove one or multiple columns from pandas dataframe.

What is pandas in Python?

pandas is a python package for data manipulation. It has several functions for the following data tasks:

1. Drop or Keep rows and columns
2. Aggregate data by one or more columns
3. Sort or reorder data
4. Merge or append multiple dataframes
5. String Functions to handle text data
6. DateTime Functions to handle date or time format columns

Import or Load Pandas library

To make use of any python library, we first need to load them up by using import command.

import pandas as pd
import numpy as np

Let's create a fake dataframe for illustration

The code below creates 4 columns named A through D.

df = pd.DataFrame(np.random.randn(6, 4), columns=list('ABCD'))

          A         B         C         D
0 -1.236438 -1.656038

String Functions in Python with Examples

This tutorial outlines various string (character) functions used in Python. To manipulate strings and character values, python has several in-built functions. It means you don't need to import or have dependency on any external package to deal with string data type in Python. It's one of the advantage of using Python over other data science tools. Dealing with string values is very common in real-world. Suppose you have customers' full name and you were asked by your manager to extract first and last name of customer. Or you want to fetch information of all the products that have code starting with 'QT'.

Table of Contents

---

List of frequently used string functions The table below shows many common string functions along with description and its equivalent function in MS Excel. We all use MS Excel in our workplace and familiar with the functions used in MS Excel. The comparison of string functions in MS EXCEL and Python would help you to learn

The cost of an open source contribution

TDK-Micronas partners with Quansight to sponsor Spyder

This blogpost was originally published on the Quansight Labs website

TDK-Micronas is sponsoring Spyder development efforts through Quansight Labs. This will enable the development of some features that have been requested by our users, as well as new features that will help TDK develop custom Spyder plugins in order to complement their Automatic Test Equipment (ATE’s) in the development of their Application Specific Integrated Circuits (ASIC’s).

At this point it may be useful to clarify the relationship the role of Quansight Labs in Spyder's development and the relationship with TDK. To quote Ralf Gommers (director of Quansight Labs):

"We're an R&D lab for open source development of core technologies around data science and scientific computing in Python. And focused on growing communities around those technologies. That's how I see it for Spyder as well: Quansight Labs enables developers to be employed to work on Spyder, and helps with connecting them to developers of other projects in similar situations. Labs should be an enabler to let the Spyder project, its community and individual developers grow.

Why citations are not enough for open source software

Spyder 4.0 takes a big step closer with the release of Beta 2!

This blogpost was originally published on the Quansight Labs website

It has been almost two months since I joined Quansight in April, to start working on Spyder maintenance and development. So far, it has been a very exciting and rewarding journey under the guidance of long time Spyder maintainer Carlos Córdoba. This is the first of a series of blog posts we will be writing to showcase updates on the development of Spyder, new planned features and news on the road to Spyder 4.0 and beyond.

First off, I would like to give a warm welcome to Edgar Margffoy, who recently joined Quansight and will be working with the Spyder team to take its development even further. Edgar has been a core Spyder developer for more than two years now, and we are very excited to have his (almost) full-time commitment to the project.

Since August 2018, when the first beta of the 4.x series was released, the Spyder development team has been

The Role of a Maintainer

What are the expectations and best practices for maintainers of open source software libraries? How can we do this better?

This post frames the discussion and then follows with best practices based on my personal experience and opinions. I make no claim that these are correct.

First, the most common answer to this question is the following:

• Q: What are expectations on OSS maintainers?
• A: Nothing at all. They’re volunteers.

However, let’s assume for a moment that these maintainers are paid to maintain the project some modest amount, like 10 hours a week.

How can they best spend this time?

Next, let’s disambiguate the role of developer, reviewer, and maintainer

1. Developers fix bugs and create features. They write code and docs and generally are agents of change in a software project. There are often many more developers than reviewers or maintainers.

2. Reviewers are known

Using GitHub for janky project reporting - some code

For the NIH Data Commons, we needed a way for 10 distinct teams to do reporting at the level of about 50-100 milestones per team, on a monthly basis.

Each team was already using different project management software internally, and we didn't want to require them to switch to something new. We also didn't need a lot of innate functionality in the project reporting system - basically, for each milestone we needed two statuses, "started" and "finished".

So we decided to go with something lightweight and simple that would support programmatic update and automated reporting: GitHub!

We chose to use GitHub for project reporting for several reasons. We were already using GitHub for content stuff, and everyone had accounts. We were also using GitHub for authentication control on static Web sites via a Heroku app.

So what we did was use the PyGithub package to write a script to take the project milestones (which were all in a spreadsheet) and load them

My first DNF (Ft Bragg 600k)

It's been six years since my first ride with The San Francisco Randonneurs and four years since my first 200k. I've ridden 18 rides that are at least that distance since then (3x 300k, 2x 400k, 1x 600k), completing my first Super Randonneur Series (2-, 3-, 4-, and 600k in one year) last year after not riding much the year before that. And this weekend I had my first DNF result on the Fort Bragg 600k. I Did Not Finish.

The best response to my choice of abandoning the ride to enjoy the campground came from Peter Curley, who said "That was a very mature decision." A clear departure from typical randonneuring stubbornness and refusal to give up, I celebrated my decision to quit as a victory when I arrived at the campground and made my announcement to the volunteers. I think I was so energetic about it that they did not believe me. I was being kind to myself, to my body, and at peace with the decision by

Should I Resign from My Full Professor Job to Work Fulltime on Cocalc?

PyCon2019 poem

I'm back in Cleveland for another Pycon. Yesterday was my first full day here. Along with Matt Seale, I was a helper at Matthias Bussonnier tutorial ("IPython and Jupyter in Depth: High productivity, interactive Python). The sticky system is efficient at signaling when someone in a classroom needs help, and a lot of folks don't know that this practice was popularized by Software Carpentry workshops and continues to be used at The Carpentries.

I stepped out for a coffee refill and bumped into a large contingent of Bloomberg folks I'd never met (Princeton office). I guess we have something like 90 people at the conference this year, and I made the usual and true remark about how I go to conferences to meet the other people who work at our company. Then after his tutorial concluded, Matthias and I bumped into Tracy Teal, exchanged some stickers, and chatted about The Carpentries, Jupyter, organizing conferences, governance and sponsorship models, and a bunch of other stuff.

Matthias was a

Why you should cite open source tools

Python Lambda Function with Examples

This article covers detailed explanation of lambda function of Python. You will learn how to use it in real-world data scenarios with examples.

Table of Contents

---

Introduction : Lambda FunctionIn non-technical language, lambda is an alternative way of defining function. You can define function inline using lambda. It means you can apply a function to some data using a single line of python code. It is called anonymous function as the function can be defined without its name. They are a part of functional programming style which focus on readability of code and avoids changing mutable data.

Syntax of Lambda Function

lambda arguments: expression

Lambda function can have more than one argument but expression cannot be more than 1. The expression is evaluated and returned. Example

addition = lambda x,y: x + y
addition(2,3) returns 5

In the above python code, x,y are the arguments and x + y is the expression that gets evaluated and returned.

Loops in Python explained with examples

Some questions and thoughts on journal peer review.

I just had the dispiriting experience of receiving a paper to review from Journal B, that was unchanged from a prior submission to Journal A. The "dispiriting" part of the experience was that the paper was completely unchanged, despite a host of minor and major comments on the paper from all three reviewers for Journal A.

I ended up writing that I was disappointed that the authors had not seen fit to confront the bigger issues in any way, much less correct even the smallest and easiest errors; and then pasted in my previous review. What I wanted to do was paste in the expert reviews from the other two reviewers for Journal A, but I didn't feel like that was OK.

(If I get the paper back with some revisions, I'll reevaluate it in light of the Journal A reviews, too.)

I think the behavior of

Create Dummy Data in Python

import pandas as pd
data = pd.DataFrame({"A" : ["John","Deep","Julia","Kate","Sandy"], 
                     "MonthSales" : [25,30,35,40,45]})

       A  MonthSales
0   John          25
1   Deep          30
2