Kaggle: Data Visualization

• This course teaches how to leverage Seaborn to make a useful yet simple way to make a data visualization.
• Start with making a line chart with a command sns.lineplot(data=the_dataset_name). sns stands for an alias of Seaborn library.
• A good to know part here is the spotify_data = pd.read_csv(spotify_filepath, index_col="Date", parse_dates=True), it’s good to know that we can consider define a column be the index and parse a column if we have a date data type (imo, this can be a wise data cleaning approach).
• Next, we learn Bar chart and Heatmaps (chart), it has different ways to write the code. The Bar chart uses sns.barplot(x=flight_data.index, y=flight_data['NK']), the x-axis and y-axis needs to be defined. While the Heatmaps, sns.heatmap(data=flight_data, annot=True), similar to the Line chart, with additional annot argument for displaying the value (number) of the data on each cell.
• For the Scatter plot, it becomes more advanced. It has similar techniques with the Bar chart, defining x and y axis, but instead using .index as the x-axis, it can be more flexible (at least by the course) sns.scatterplot(x=insurance_data['bmi'], y=insurance_data['charges'])
• We also learn how to make regression plot, sns.regplot(x=insurance_data['bmi'], y=insurance_data['charges']), just add the line of code below the scatter plot chart so it can combined together.
• Figure below shows a use of 3 variables: bmi, charges, and smoker. It applies on sns.scatterplot() to show relationship of 3 variables.

• This is the explanation of above plot, This scatter plot shows that while nonsmokers to tend to pay slightly more with increasing BMI, smokers pay MUCH more.

• There’s more plot to mention such as sns.lmplot and sns.swarmplot which has different approaches for visualizing data.

• Histogram, is’t same with Bar plot. It uses a range of value on the x-axis which makes the bar-graph doesn’t exactly on the category axis.

• Kernel Density Estimate (KDE), is a smoothed histogram

• Different types of data categories:
  • Trends - A trend is defined as a pattern of change.
  • Relationship - Between one and another features
  • Distribution
• Not that important, but good to know. sns.set_style("dark") a command to give a style to our graph/plot. Put or define the code before we make our plot.