But we can do better, right? Of course we can! Let’s add more variables to the model. We get a model with a mean squared error of 28.66 and an R² of 0.59. Print("Mean squared error: %.2f" % mean_squared_error(y_test, y_predicted)) Some good evaluation metrics for linear regression are mean squared error and the R² score. In the first instance, we run the model on our test set. Now that we have our model, we can check how well it performs. We want to predict the mileage per gallon by looking at the horsepower of a car. First, let’s try a model with only one variable. X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,įinally, we can start building the regression model. Scikit-learn has a very straightforward train_test_split function for that.
df = df.drop(, axis=1)įinally, we’ll split the dataset into a train set and a test set. Finally, the ‘mpg’ column is dropped in the X variable and set as the target in the Y variable. These columns are the model name, the geographical origin and the year that the model was built. In the following part, for educational purposes, we’ll drop some columns that I don’t think we need in our regression model. Using pandas, we replace question marks with NaNs and remove these rows. Next up, we will clean the dataset and remove the missing values.
An introduction to statistical learning [df download#
You can download the famous mpg dataset from the UCI Machine Learning Repository, or just google “mpg.csv.” Using pandas, you can quickly read in the CSV into a DataFrame. import pandas as pdįrom sklearn.linear_model import LinearRegressionįrom sklearn.model_selection import train_test_splitįrom trics import mean_squared_error, r2_score In this tutorial, I will briefly explain doing linear regression with Scikit-Learn, a popular machine learning package which is available in Python. Truth be told, if you’re interested in all the mathematical details of linear regression (which I strongly recommend learning about), get an econometrics book. The slope is the marginal effect of increasing X by one unit. The intercept is the value of your prediction when the predictor X is zero. There are two coefficients in this model: the intercept and the slope. We assume a linear relationship between the quantitative response Y and the predictor variable X. Simple linear regression is pretty straightforward. In “An introduction to Statistical Learning,” the authors claim that “the importance of having a good understanding of linear regression before studying more complex learning methods cannot be overstated.” Moreover, it is the origin of many machine learning algorithms. Linear regression is one of the most popular techniques for modelling a linear relationship between a dependent and one or more independent variables.
0 kommentar(er)
