Visualization of Cleaned Data: Horsepower vs. Miles per Gallon
Now that the dataset is cleaned and available as cleanedData, we can visualize the relationship between Horsepower and Miles_per_Gallon. This scatter plot will help us understand the data points that will be used for training our model.
Splitting the Dataset: Training and Testing Sets
To evaluate a machine learning model fairly, it's important to train it on a portion of the data and test it on data it hasn't seen. The dataset is typically split into a training set (for model learning) and a testing set (for evaluation). A common split is 70% for training and 30% for testing.
Building an Artificial Neural Network Model
Neural networks are powerful tools for modeling complex nonlinear relationships in data. Here, we’ll create a simple artificial neural network (ANN) that learns to predict Miles_per_Gallon from Horsepower using the training data we prepared earlier.
What is an Artificial Neural Network?
An artificial neural network (ANN) is inspired by biological brains. It consists of interconnected layers of simple processing units called neurons. Each neuron receives inputs, applies weights, computes a sum, and then applies an activation function to determine its output.
- The input layer receives features (here, horsepower).
- The hidden layer processes the input through several neurons.
- The output layer produces the final prediction (here, miles per gallon).
An example of a simple neural network with an input, a hidden, and an output layer.
In this project, the model will learn from the training set you created to estimate the relationship between horsepower and miles per gallon, aiming to predict efficient vehicle performance.
Training the Neural Network Model
Training means letting the neural network learn patterns from data. Here, the model will learn to predict Miles_per_Gallon from Horsepower using the cleanedData you loaded previously.
Evaluating the Trained Neural Network Model
Once the model is trained, it’s important to check how well it performs on new data it hasn't seen before. This is done using the test set that was separated from the cleanedData. Here, we’ll compare the model’s predictions with the real values to understand its performance.
Understanding the Confusion Matrix
The confusion matrix is a useful tool for evaluating classification models. It shows the number of correct and incorrect predictions made by the model compared to the actual outcomes (labels). The matrix helps you see not only the overall accuracy, but also the types of errors the model makes.
- True Positive (TP): Model correctly predicts the positive class.
- True Negative (TN): Model correctly predicts the negative class.
- False Positive (FP): Model incorrectly predicts positive when it is actually negative (Type I error).
- False Negative (FN): Model incorrectly predicts negative when it is actually positive (Type II error).
A confusion matrix for a binary classifier looks like this:
Actual
| 1 | 0
---+---+---
1 | TP | FP
Pred. ---+---+---
0 | FN | TN
To use a confusion matrix with this regression demo, we will convert the regression output into two classes: for example, cars with MPG above 23 are "Efficient" (1), and those with MPG 23 or below are "Not Efficient" (0). The threshold can be adjusted as needed.