I am working on machine learning model who will predict Whether the team will won or not. but there are one of more important features spread point in NFL Football Games , Can anyone know ? How i predict it and what are the best features for the this feature? I am adding dataset here link . Here are the features avaible for the predicting results in result column [L,W,D] lose,won,draw respectivly. But need help how to predict spread point ?
1 Answers
The machine learning model needs to be a binary classifier and designing a prediction model often requires trying multiple models with different parameter settings and comparing the accuracies of each model on the same dataset. You can find a list of such classifiers to experiment with at Supervised Classifiers.
There are various Feature Selection techniques to identify important features for a dataset, including manual feature extraction, but a technique you could try out on the NFL dataset is called ensemble classifier, specifically Random Forest Classifier. There are other techniques one should definitely explore, but the Random Forest Classifier of sklearn RandomForestClassifier returns an attribute called feature_importances_ which is an array of values between 0 to 1 indicating the importance of each feature, these can also be plotted as a graph.
It is important to fit the model on the dataset, using sklearn classifier fit() method to get this list of feature importance, you could then identify which features to use based on these values. The Random Forest classifier itself will also be able to provide predictions, but it is important to try out different classifiers.
It should be noted that the feature_importances_ provided by RandomForestClassifier is not a universal solution for feature selection. The choice of features and classifier heavily depends upon the data itself, thus, initial data analysis is always recommended for any dataset to identify if the data has a high bias or high variance and identify correlation among different features.
Related