Data Analysis From Scratch With Python: Step By Step Guide

y_pred = regressor.predict(X_test)
y_pred (predicted Profit values on the
X_test) will be like this: 
However, is that all there is? Are all the variables (R&D Spend, Administration,
Marketing Spend, State) responsible for the target (Profit). Many data analysts
perform additional steps to create better models and predictors. They might be
doing Backward Elimination (e.g. eliminating variables one by one until there’s
one or two left) so we’ll know which of the variables is making the biggest
contribution to our results (and therefore more accurate predictions).
There are other ways of making the making the model yield more accurate
predictions. It depends on your objectives (perhaps you want to use all the data
variables) and resources (not just money and computational power, but also time
constraints).
Decision Tree
The Regression method discussed so far is very good if there’s a linear
relationship between the independent variables and the target. But what if there’s
no linearity (but the dependent variables can still be used to predict the target)?
This is where other methods such as Decision Tree Regression comes in. Note
that it sounds different from Simple Linear Regression and Multiple Linear
Regression. There’s no linearity and it works differently. Decision Tree
Regression works by breaking down the dataset into smaller and smaller subsets.
Here’s 
an 
illustration 
that 
better 
explains 
it: 

---

http://chem-eng.utoronto.ca/~datamining/dmc/decision_tree_reg.htm
Instead of plotting and fitting a line, there are decision nodes and leaf nodes.
Let’s quickly look at an example to see how it works (using
Position_Salaries.csv): The dataset: 
Position,Level,Salary
Business Analyst,1,45000