ValueError: Expected 2D array, got 1D array instead (Python) — how to fix
ValueError: Expected 2D array, got 1D array instead:
array=[1. 2. 3. 4. 5.].
Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample.
Quick fix: Single feature column: X = X.reshape(-1, 1) or use df[["col"]] (double brackets) instead of df["col"]. Single sample: model.predict([[5.0, 3.2]]) or x.reshape(1, -1). Keep the target y 1-D.
What it means
Every scikit-learn estimator (fit, predict, transform, score) takes a 2-D feature matrix X: one row per sample, one column per feature. A 1-D array is ambiguous: it could be five samples with one feature, or one sample with five features. sklearn refuses to guess and tells you which reshape to use.
import numpy as np
x = np.array([1, 2, 3, 4, 5])
x.shape # (5,) 1-D
x.reshape(-1, 1).shape # (5, 1) 5 samples, 1 feature
x.reshape(1, -1).shape # (1, 5) 1 sample, 5 features
-1 means "work this dimension out from the array size".
Cause 1: a single feature column
from sklearn.linear_model import LinearRegression
X = df["area"] # Series, 1-D
y = df["price"]
LinearRegression().fit(X, y) # ValueError
Select with a list of columns to get a DataFrame (2-D):
X = df[["area"]] # shape (n, 1)
model = LinearRegression().fit(X, y)
Or with NumPy: X = df["area"].to_numpy().reshape(-1, 1).
Cause 2: predicting a single sample
model.predict([5.1, 3.5, 1.4, 0.2]) # 1-D: ValueError
model.predict([[5.1, 3.5, 1.4, 0.2]]) # 2-D: one row
With an array:
sample = np.array([5.1, 3.5, 1.4, 0.2])
model.predict(sample.reshape(1, -1))
If the model was fitted on a DataFrame, predict with one too to avoid the X does not have valid feature names warning:
model.predict(pd.DataFrame([[5.1, 3.5, 1.4, 0.2]], columns=X.columns))
Cause 3: scalers and encoders on one column
StandardScaler, MinMaxScaler, OneHotEncoder and SimpleImputer also need 2-D input:
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
df["age_scaled"] = scaler.fit_transform(df[["age"]]).ravel()
.ravel() turns the (n, 1) result back into 1-D for assigning to a column. LabelEncoder is the exception: it is meant for the target and takes 1-D.
Cause 4: a loop or indexing removed a dimension
X[0] or X[:, 0] drops a dimension. Keep it with a slice:
X[0:1] # first row, shape (1, n_features)
X[:, [0]] # first column, shape (n_samples, 1)
Don't reshape the target
y should stay 1-D for most estimators. Passing a (n, 1) column vector as y gives DataConversionWarning: A column-vector y was passed when a 1d array was expected. Fix with y.ravel() or df["price"] (single brackets).
Last updated 2026-11-21.