cas-pml/SL/aufgaben/template/4_WS/WS 13 Vorlage.ipynb

WS 13 Kreuzvalidierung

• vergleichen Sie alle bisher bekannten Klassifikatoren (ausser SVC und MLPClassifier) in Bezug auf deren Stabilität unter Anwendung von Kreuzvalidierung
• verwenden Sie für die Klassifikatoren jeweils Default-Parametrisierung
• setzen Sie für die Kreuzvalidierung folgende Funktion ein: sklearn.model_selection.cross_val_score

In [3]:

## load libraries
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns; sns.set()

## load data
datapath = '../3_data'
from os import chdir; chdir(datapath)
bank_df = pd.read_csv('bank_data_prep.csv')

## features - target - tplit
X = bank_df.drop('y', axis=1)
y = bank_df['y']

## train - test - split
## obsolete here, is done internally by cross validation

In [4]:

from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
## tbd complete


from sklearn.linear_model import LogisticRegression

from sklearn.model_selection import cross_val_score

models = [
    KNeighborsClassifier(),
    DecisionTreeClassifier(),
    RandomForestClassifier()
    ## tbd complete
    
    
]

kfold = 5
model_names = []
model_scores = []

for model in models:
    model_name = model.__class__.__name__
    print(model_name)
    ## tbd

    
    

KNeighborsClassifier
DecisionTreeClassifier
RandomForestClassifier

In [5]:

## manage results, e.g. in pandas dataframe
## tbd



## visualize results
## tbd

Fazit:

• tbd