error: "ValueError: could not convert string to float: ''" in pandas

Question

Hello everyone for school i need to make a machine learning project where i have to predict the rainfall

my excell looks like this:

and this is my code:

import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn import metrics

df = pd.read_excel('Month.xlsx', usecols='W, AJ')
data = pd.read_excel('Month.xlsx')

df

x = df.iloc[:,0]
y = df.iloc[:,1]



model = LinearRegression()

model.fit(x, y)

y_pred = model.predict(x)

# Visualisatie van de regressielijn
plt.scatter(x, y,  color='gray')
plt.plot(x, y_pred, color='red', linewidth=2)
plt.xlabel('Luchtvochtigheid')
plt.ylabel('Regenval')
plt.show()

but i keep getting an error with "ValueError: could not convert string to float: ''. This error is given on the line where i try to model.fit(x,y)

I have tried to change the x and y values to a float with a new code like Xdata = float(x), but then i got a whole different error: cannot convert the series to <class 'float'>

Answer 1

This means that in your df you have some observations that are empty strings like this: '' that of course cannot be converted to a number.

In this case you have to decide what's the best option between replace them with a number (like zero, the mean, the mean divided by the standard deviation etc) or just drop the rows with the empty string / None values.

Before implementing an ML algorithm you should always do some data preprocessing (that is also the difficult part in a data science project), that is totally missing in your code.

Also, you are not splitting X and y into train and test, hence you are using for predictions the same data you used for training, that is an error, since the model will be completely useless.

I suggest you to take a look at some documentation on how to implement a data science project, because your approach is lacking a lot of basic stuff.