And apparently there was! Library imblearn is dedicated to help you solve this issue! You can choose different methods o

This one is a library for doing ML, similar to sklearn. Though it has UI, on which you can do your ML by clicking multiple mouse buttons. Import data, explore it, impute missing data if you’d like, and then even run AutoML, which will search on models and configuration to find the best one (also stacks them at the end!). I’d say it can help even a person having almost none programming experience to do ML.

I’ve covered how to explore and analyze your data using seaborn library, how to spot outliers and how to clean your data. Though keep in mind each machine learning task is different, and it might. Next, we checked several scikit-learn transformers and how they stack up and process your data. Following with, data class imbalance issues with imblearn library and some libraries I played around with. Added some tools and platforms which might make your life easier tracking your model performance or training them on GPUs (if they have this option).

Like I said, I did this out of curiosity; my submission to the task was not the best of what I could have achieved (only 0.78 AUC). If you wouldn’t make the same mistakes as I did, it will benefit you as well as it did for me.

This one is aimed to let you run flows using RAPIDS. You can leverage GPUs and install any additional libraries you need. The interface is Jupyter Hub. Honestly, I somehow liked Colab more than BlazingSQL. It’s a good alternative to do the same GPU accelerated computation as well. You can mount the same google drive if you want, but if you’d upload your files here, it would stay there even if your server is restarted, not like in Colab.

This snippet would fix imbalances, remove outliers and fit multiple models with a search on hyperparameters. My data was around 1GB CSV file, so even 5 folds take some time on one model. Keeping in mind that there is hyperparameter tunning multiple models 5 folds = too long waiting time. If I’d have smaller datasets, I’d definitely give it a try again; it seems easy to work with, and it does all the work for you!

Now your data preparation will be robust, and you will be sure that your train dataset and the one you’ll be basing your predictions will follow the same manipulations in the same order!

Their framework is easy to work with, specifying layers similar to MLPClassifier in sklearn — pass a list with neuron count in each layer. Though running your training on CPU will drive you mad (it takes A LOT of time, like literally A LOT). In my case, when I used the CPU, it showed 150 hours. None of my owned machines had an Nvidia GPU inside, so it wasn’t on my list to try first. In later parts, I’ll tell how to use free GPU on some platforms and speed up any model training that can leverage GPU.

I was not too fond of that H2OFrame doesn’t quite work like pandas/spark/dask data frames sometimes. It was a bit hard to figure out how to do certain things, i.e., get a list of distinct column values. Here I had to use flatten, though in pandas would be list(). Nonetheless pretty cool library which can help anyone!

and access your data in the usual way, just from /content/drive. Though I guess you agree with google seeing your data, which might not work for sensitive information. Though for Kaggle and playing around, it works quite well.

http://old.cocir.org/media/fxa/video-van-buyuksehir-belediyespor-v-eyupspor-v-tr-tr-1rev-27.php

http://old.cocir.org/media/fxa/videos-van-buyuksehir-belediyespor-v-eyupspor-v-tr-tr-1hmx-5.php

http://old.cocir.org/media/fxa/Video-van-buyuksehir-belediyespor-v-eyupspor-v-tr-tr-1zet-9.php

http://agro.ruicasa.com/vtm/videos-sariyer-v-corum-belediyespor-v-tr-tr-1say-18.php

http://agro.ruicasa.com/vtm/v-ideos-sariyer-v-corum-belediyespor-v-tr-tr-1uxg-13.php

http://m.dentisalut.com/omy/Video-utas-usakspor-v-niğde-anadolu-v-tr-tr-1cia-26.php

http://agro.ruicasa.com/vtm/videos-sariyer-v-corum-belediyespor-v-tr-tr-1ngg-16.php

http://m.dentisalut.com/omy/v-ideos-utas-usakspor-v-niğde-anadolu-v-tr-tr-1nqq-20.php

http://agro.ruicasa.com/vtm/v-ideos-sariyer-v-corum-belediyespor-v-tr-tr-1ydl-2.php

http://old.cocir.org/media/fxa/video-sakaryaspor-v-kastamonuspor-v-tr-tr-1ama-13.php

http://m.dentisalut.com/omy/v-ideos-utas-usakspor-v-niğde-anadolu-v-tr-tr-1bee-10.php

http://m.dentisalut.com/omy/videos-utas-usakspor-v-niğde-anadolu-v-tr-tr-1sek-9.php

http://old.cocir.org/media/fxa/videos-sakaryaspor-v-kastamonuspor-v-tr-tr-1mii-25.php

http://agro.ruicasa.com/vtm/video-pendikspor-v-bayburt-ozel-v-tr-tr-1bse-7.php

http://old.cocir.org/media/fxa/videos-sakaryaspor-v-kastamonuspor-v-tr-tr-1xjt-14.php

http://agro.ruicasa.com/vtm/v-ideos-pendikspor-v-bayburt-ozel-v-tr-tr-1xxm-11.php

http://m.dentisalut.com/omy/videos-ağri-1970-spor-v-ceyhan-v-tr-tr-1dge-1.php

http://m.dentisalut.com/omy/v-ideos-ağri-1970-spor-v-ceyhan-v-tr-tr-1hvr-6.php

http://agro.ruicasa.com/vtm/v-ideos-pendikspor-v-bayburt-ozel-v-tr-tr-1hwb-13.php

http://old.cocir.org/media/fxa/video-sakaryaspor-v-kastamonuspor-v-tr-tr-1lgw-21.php

unning our own studio we have hired exactly one Black person. Sentence about the pipeline yadda yadda yadda. Sentence about how you can only hire who applies yadda yadda yadda. Once all the excuses that you attempt to use to cover your shame are exhausted you’re left with the truth: we have hired exactly one Black person in twenty years. That’s a fact. And it’s a fact that makes me very uncomfortable, because uncomfortable is where we need to be. It’s also a fact that most of the companies in our industry are no better. You have not hired enough Black people. And if reading this is making you uncomfortable, great.

Beware of some limitations. Collaboratory was designed for people to test some things, get familiar and share at ease. So probably, you won’t be able to run long pipelines on the free version. This version gives 12h of being connected (after that, you can’t use GPU). Also, it timeouts after 90mins of nothing being pressed, basically disrupting complex, long-running neural nets or computations; you have to click some things here and there to avoid it).

How many times you ran a model, got a good score, but the next day you couldn’t reproduce because you’ve made some changes here and there (different configuration, dataset)? At least to me, it was getting out of hand quite fast. Only printing some configuration, and the AUC score was getting lost pretty fast. I needed to track all of the configurations, scores, and data in general somehow. Using this platform, I could have logged my best model and then fine-tune it. Best discovery for me, though it was a bit too late for the challenge.

MLflow — A platform for the machine learning lifecycle An open-source platform for the machine learning lifecycle MLflow is an open-source platform to manage the ML…mlflow.org.

Let’s do a hyperparameter tuning for them. Usual helpers are GridSearchCV (exhaustive search) or RandomSearchCV (not necessarily get the best combination) from sklearn. I was so glad that I found out about this amazing gem — scikit-optimize. It has BayesSearchCV. It takes ranges and takes a parameter of several iterations. It should try to optimize your model based on the scorer you provide. Minor issue with this that it could fall to a local minimum, and you wouldn’t improve anywhere.

The most common (or at least it was for me) understanding that sklearn is the most used library for ML. Before going to baseline results, I’m going to present some of the libraries. Keep in mind I’m aiming here at fresh Data Scientist or this path explorers, so I’m not going to Neural Network libraries (Keras, Tensorflow, PyTorch).