Photo by Luca Bravo on Unsplash

Timeseries can be hard. Timeseries may require a lot of feature engineering to get the job done. And even then the results may seem a bit underwhelming with respect to the complexity put into the model. In this blogpost, I want to show how timeseries can be approached with 1D Convolutional Neural Nets and how impressive the results are. So buckle up for the ride!

For this post, I will use the Italy Power demand dataset. The classification task is to distinguish days from Oct to March (inclusive) from April to September. The dataset can be found here. The best…


Photo by Mika Baumeister on Unsplash

When it comes to DeepLearning, the more data we have the better the chances are to get a great performing model. In fields like image recognition research has already came up with quite a few clever ideas how to use the existing data to create more data out of it. This is called data augmentation.

However, when we look at Deep Learning in the tabular data context, there are still many concepts missing. …


Photo by Jorge Ramirez on Unsplash

When it comes to highly imbalanced data and how to get better results on the almost non-existent target class, “classic” machine learning algorithms already came up with a few ideas. A classic one is oversampling of the underrepresented class. In sklearn for example you can give class weights to each class and thereby try to better fit your model to the underrepresented class.

Other techniques like SMOTE are based on the idea that you can create “new” data for those underrepresented classes and thereby make better predictions. The idea I want to present here has a similar approach of how…


Making pystan run on your computer can and probably will be a hassle, which might end up not using pystan anyway. I came across quite a few obstacles in the process of making it work and I would like to guide you how to make pystan run on your computer (yes, even on a Windows!).

First of all you need to check the requirements pystan needs in order to run:

  • Python ≥3.7
  • Linux or macOS
  • x86–64 CPU
  • C++ compiler: gcc ≥9.0 or clang ≥10.0.

First things first, I made pystan run under Python 3.7.10 and Python 3.8.5 on Ubuntu 20.04…


Photo by Daniel Cheung on Unsplash

Have you ever wonder how to make practical use of artifical intelligence in your personal life? Look no further! In this blogpost I will show you how to download a dataset from kaggle from within your jupyter notebook and use fastai to finally solve the problem of correctly classifying Lego figurines.

To download datasets directly from kaggle we need to make use of the kaggle API. This basically means you have to log in into your kaggle account, go to “Your Account” and scroll down until you find the section API. Click on “Create New API Token” and put it…


Photo by fabio on Unsplash

Motivation

Most of the talk in Deep Learning is on either NLP or image recognition. These are all interesting cases in and by itself, however most of the business cases usually do not revolve around these topics. You can use many of the techniques already established from NLP or image recognition, still there are a few differences on how to better make use of Deep Learning on tabular data. This article should stress out these differences and share some insights on what works for Deep Learning on tabular data and what not.

In my experience a great deal of business-relevant cases…

Lasse Schmidt

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store