Summary

Quite a long chapter! Isn't it? But, this chapter will form the core of anything you learn and implement in data-science. Let us wrap-up the chapter by summarizing the key takeaways from the chapter:

  • Data can be sub-setted in a variety of ways: by selecting a column, selecting few rows, selecting a combination of rows and columns; using .ix method and [ ] method, and creating new columns.
  • Random numbers can be generated in a number of ways. There are many methods like randint(), raandarrange() in the random library of numpy. There are also methods like shuffle and choice to randomly select an element out of a list. Randn() and uniform() are used to generate random numbers following normal and uniform probability distributions. Random numbers can be used to run simulations and generate dummy data frames.
  • The groupby() method creates a groupby element on which aggregate, transform, and filter operations can be applied. This is a good method to summarize data for each categorical variable at once.
  • A data must be split between training and testing datasets before a modelling is performed. The training dataset is the one on which the model equations are developed. The testing dataset is used to test the performance of the model comparing the actual result (present in testing dataset) to the model output. There are various ways to perform this split. One can use choice and shuffle. Scikit-learn has a readymade method for this.
  • Two datasets can be merged just like two tables in a relational database. There are various kind of joins—Inner, Left, Right, Outer, and so on. These joins can be understood better if the datasets are assumed analogous to sets. Inner Join is then Intersection, Outer Join is Union, and Left and Right joins are entire left and right data frame.

Wrangling data and bringing it in the form you desire is a big challenge before one proceeds to modelling. But, once done, it opens up a plethora of insights and information to be discovered using predictive models. As Bob Marley said, "If it is easy, it won't be amazing; if it is amazing, it won't be easy."

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset