<script async src="https://www.googletagmanager.com/gtag/js?id=UA-123065321-1"></script> <script> window.dataLayer = window.dataLayer || []; function gtag(){dataLayer.push(arguments);} gtag('js', new Date()); gtag('config', 'UA-123065321-1'); </script>

Welcome!

This is a place for me to share Data Science resources where I aim to provide lots of simple python examples directly in Google Colab.

---

Notebooks

Pandas I

Basic pandas operations
Build dataframes
Modify series
Merge and concat

Pandas II

Handle missing data
Convert types
Group by
Aggregate
DateTime

Plotting Data

Scatter plots for EDA
K-means clusters
Seaborn heatmap
Seaborn distplots
Pandas boxplot
Scipy interpolate
Scipy Q-Q plots

Miscellaneous

Setup a class framework for use in notebooks, part 1 of 2
Setup a class framework for use in notebooks, part 2 of 2
Process text files
Extracting table data from the web
List comprehensions
Numpy basics
Precision-Recall vs ROC curves

---

Machine Learning - Example walkthroughs

Breast cancer prediction - python notebook that demonstrates the following techniques:

1. Univariate feature reduction (remove low correlations with the target).
2. Feature reduction based on collinearity (for each highly correlated pair of features, leave only the feature that correlates better with the target value).
3. Compare two different model types for supervised learning (Logistic Regression and GBM), including the testing and ranking of feature importance.
4. Calculate percentile bins for each model in order to determine the ratio of positive classes for each percentile bin.
5. Plot ROC and Precision-Recall curves.

TF-IDF based topic clustering using PCA with K-means, NMF, LDA - python notebook that demonstrates the following techniques:

1. Vectorize text to a numeric matrix using TF-IDF (Term Frequency - Inverse Document Frequency)
2. Dimensionality Reduction using PCA
3. Unsupervised classification: Calculate K-means clusters based on PCA (a reduced version of TF-IDF)
4. Unsupervised classification: Calculate NMF (Non-negative Matrix Factorization) based on TF-IDF
5. Unsupervised classification: Calculate LDA (Latent Derilicht Analysis) based on TF

Article for the above python notebook:

Comparing the performance of non-supervised vs supervised learning methods for NLP text classification

---

Connect on LinkedIn

Copyright © Gal Arav, 2020