从 numpy arrays 创建 dataframe

Question

I have the following code and I am stuck in creating a pandas dataframe by some numpy arrays.

def gradient_descent_classification_prediction(x_test):
    y_pred = pd.DataFrame()
    for i in range(0 , 10):
        print(i)
        b = pd.read_csv("theta"+str(i)+".csv" , delimiter=",")
        y = pred(x_test , b) #outputs are 2d-s numpy arrays (10000 , 1) 
        y_pred= y_pred.append(y.tolist()) 
    y_pred.to_csv("./y_pred.csv" , index=False)

I would like to have a column in the dataframe for each y which is calculated by the function pred(x_test, b)

The output that I get is a single column dataframe with just an output of the function pred.

Could you help me?

Answer 1

There are many ways to do this but essentially append a list to dataframe does not ensure its shape. It's easier to keep it as a dictionary or list, then convert to data frame before writing. Below I use a dictionary:

def gradient_descent_classification_prediction(x_test):
    y_pred = {}
    for i in range(0 , 10):

        b = pd.read_csv("theta"+str(i)+".csv" , delimiter=",")
        y = pred(x_test , b) #outputs are 2d-s numpy arrays (10000 , 1) 
        y_pred["theta"+str(i)] = y
    
    pd.DataFrame(y_pred).to_csv("./y_pred.csv" , index=False)