Python RandomForest - Unknown label Error

Question

I have trouble using RandomForest fit function

This is my training set

         P1      Tp1           IrrPOA     Gz          Drz2
0        0.0     7.7           0.0       -1.4        -0.3
1        0.0     7.7           0.0       -1.4        -0.3
2        ...     ...           ...        ...         ...
3        49.4    7.5           0.0       -1.4        -0.3
4        47.4    7.5           0.0       -1.4        -0.3
... (10k rows)

I want to predict P1 thanks to all the other variables using sklearn.ensemble RandomForest

colsRes = ['P1']
X_train = train.drop(colsRes, axis = 1)
Y_train = pd.DataFrame(train[colsRes])
rf = RandomForestClassifier(n_estimators=100)
rf.fit(X_train, Y_train)

Here is the error I get:

ValueError: Unknown label type: array([[  0. ],
       [  0. ],
       [  0. ],
       ..., 
       [ 49.4],
       [ 47.4],

I did not find anything about this label error, I use Python 3.5. Any advice would be a great help !

Answer 1

When you are passing label (y) data to rf.fit(X,y), it expects y to be 1D list. Slicing the Panda frame always result in a 2D list. So, conflict raised in your use-case. You need to convert the 2D list provided by pandas DataFrame to a 1D list as expected by fit function.

Try using 1D list first:

Y_train = list(train.P1.values)

If this does not solve the problem, you can try with solution mentioned in MultinomialNB error: "Unknown Label Type":

Y_train = np.asarray(train['P1'], dtype="|S6")

So your code becomes,

colsRes = ['P1']
X_train = train.drop(colsRes, axis = 1)
Y_train = np.asarray(train['P1'], dtype="|S6")
rf = RandomForestClassifier(n_estimators=100)
rf.fit(X_train, Y_train)

Answer 2

According to this SO post, Classifiers need integer or string labels.

You could consider switching to a regression model instead (that might better suit your data, as each datum appears to be a float), like so:

X_train = train.drop('P1', axis=1)
Y_train = train['P1']
rf = RandomForestRegressor(n_estimators=100)
rf.fit(X_train.as_matrix(), Y_train.as_matrix())

Answer 3

may be a tad late to the party but I just got this error and solved it by making sure my y variable was type(int) using

 y = df['y_variable'].astype(int) 

before doing a train test split, also like others have said you problem seems better fit with a RFReg rather then RF