RandomForest Module

A random forest model.

The doc is here : http://scikit-learn.org/dev/modules/generated/sklearn.ensemble.RandomForestClassifier.html

C’est le premier modèle que j’ai utilisé. Pour 5 arbres, Kaggle me donne 75.59% de réussites.

---

Sortie du script

$ python RandomForest.py
Opening the file 'train.csv' and 'test.csv'...
Find the best value for the meta parameter n_estimators, with 10 run for each...
Searching in the range : [1, 2, 4, 5, 8, 10, 20]...
Using the first part (67.00%, 596 passengers) of the training dataset as training, 
and the second part (33.00%, 295 passengers) as testing !
For 1 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 84.09%...
For 2 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 84.03%...
For 4 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 87.40%...
For 5 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 87.87%...
For 8 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 89.24%...
For 10 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 88.71%...
For 20 random tree(s), learning from the first part of the dataset...
... this value of n_estimators seems to have a (mean) quality = 90.25%...
With trying each of the following n_estimators ([1, 2, 4, 5, 8, 10, 20]), each 10 times, the best one is 20. (for a quality = 90.25%)
Creating the random forest (of 20 estimators)...
Learning...
 Proportion of perfect fitting for the training dataset = 97.31%
Predicting for the testing dataset
Prediction: wrote in the file csv/RandomForest_best.csv.

Résultats

La soumission du résultat à Kaggle donne 77.38%.

---

RandomForest.list_n_estimators = [1, 2, 4, 5, 8, 10, 20]

Espace de recherche

RandomForest.Number_try = 10

Nombre de tests utilisés pour méta-apprendre

RandomForest.proportion_train = 0.67

Proportion d’individus utilisés pour méta-apprendre.

RandomForest.best_n_estimators = 20

La valeur optimale trouvée pour le paramètre n_estimators

RandomForest.score = 97.194163860830528

The score for this classifier.