Day #80 - Visualizations

EDA is an art. Visualizations are art tools. Several different plots to prove hypothesis

Visualization Tools

• Histograms (Split into bins, how many points fall in each bins, vary number of bins) - plt.hist(x)
• XGBoost will benefit from explicit missing values
• Plots - index versus value, plt.plot(x,'.'), randomness over indices
• Statistics

Explore Feature Relations

• Scatter Plots (Draw one features vs other), Data distribution between train and test tests validate how they are distributed
• Correlation Plots (Run K-means clustering and reorder feature) - How similar features are
• Plot (index vs feature statistics)

Feature Groups

• Generate new features based on groups

Pairs

• ScatterPlot, Scatter matrix
• Correlation Plot (Corrplot)

 Groups

•  Corrplot + Clustering
•  Plot (Index vs feature statistics)

More Read (Link)

Which color scale to use when visualizing data

Best practices for BI dashboards

Happy Learning and Coding!!!