Regression problem:

In a regression problem, the aim is to predict the output of a continuous value, like a price or a probability. Contrast this with a classification problem, where the aim is to select a class from a list of classes (for example, where a picture contains an apple or an orange, recognizing which fruit is in the picture).

Dataset:

This tutorial uses the classic Auto MPG dataset and demonstrates how to build models to predict the fuel efficiency of the late-1970s and early 1980s automobiles. To do this, you will provide the models with a description of many automobiles from that time period. This description includes attributes like cylinders, displacement, horsepower, and weight.

Project steps:

1. Data Exploration.

• The top row suggests that the fuel efficiency (MPG) is a function of all the other parameters.

2. Preprocessing: mainly cleaning and convert categorical values one hot encoding.

3. Data splitting

4. Creating the models

a. Single feature with linear regression.

The plot of the HorsePower feature with MPG.

b. Multiple features with liner regression.

c. Single feature with DNN.

The plot of the HorsePower feature with MPG.

d. Multiple features with DNN.

Performance Results

Model	Mean absolute error [MPG]
Single_feature_Regression	3.644525
linear_model_MultipInputs	3.364622
dnn_model_single_input	2.910135
dnn_mi	2.734736
reloaded	2.734736

Referances:

1. Basic regression: Predict fuel efficiency.
2. 04_regression, tensorflow-course