Home / Questions / 274135
Accepted
efueyo
Asked: 2020-06-24 10:44:36 +0800 CST 2020-06-24 10:44:36 +0800 CST

ValueError: negative dimensions are not allowed

  • 772

The Python package called Patsy helps to define a formula with dependent and independent variable. The variable defined to the left of '~'is the dependent variable, and the variables defined to the right of it are the independent variables. Variables enclosed within C() are treated as categorical variables. From the DataFrame "titanic_3.csv", I need to create two sets: training set and testing set. This DataFrame is accessible in the Files folder of this link

When executing the script

import os
import pandas as pd
from patsy import dmatrices
# Change directory 
os.chdir("G:\Py_machine_learning\Ficheros")
df = pd.read_csv("titanic_3.csv")
# Aplicar axis as 1 para eliminar las columnas con las siguientes etiquetas
df = df.drop(["ticket","cabin","name", "Unnamed: 14", "Unnamed: 15", "Unnamed: 16",
              "Unnamed: 17", "Unnamed: 18", "Unnamed: 19"], axis=1)
# Eliminamos valores Na
df = df.dropna()
from patsy import dmatrices
formula = "survived ~ C(pclass) + C(sex) + age + sibsp + C(embarked) + parch" 
# Crea un diccionario de resultados para mantener nuestros resultados de regresión 
# para un fácil análisis posterior
df_train = df.iloc[ 0: 600, : ]
df_test = df.iloc[ 600: , : ]
# Divide los datos en variables dependientes e independientes.
# Crea los conjuntos de datos de entrenamiento y prueba
y_train,x_train = dmatrices(formula, data=df_train, return_type = 'dataframe')
y_test,x_test = dmatrices(formula, data=df_test,  return_type = 'dataframe')

it returns me the error:

ValueError: negative dimensions are not allowed What is the cause of this error?

python

1 Answers

  1. Best Answer

    After doing the df.dropna()resulting dataframe only has 142 rows in this example, so when trying to select from row 600 onwards df_testyou end up with an empty dataframe, hence the error.

    It is true that the message is not intuitive, by mentioning a negative dimension that is not seen in your code, but this happens because internally the patsy tries to create a unit matrix (ones on the diagonal and zeros outside) of dimensions len(levels)-1, calling np.eye(), and since it levels's zero in this case, it ends up causing a negative dimension that causes the exception in np.eye().

    • 0