I'm trying to STOCHASTICALLY assign a fourth value (1 of 2 types of buddy) based on value of category value.
small df with randomly assigned values for 3 features: category, age and sex
Unique_ID Category Age Sex Buddy
0 0 2 11 male NaN
1 1 3 7 female NaN
2 2 1 4 male NaN
3 3 2 20 male NaN
4 4 1 19 female NaN
i include code to generate df if helpful for answer
i've made a function to hard-coding the probabilities for np.random.choice, but am running into error message when applying assign_buddy function to df ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().
columns = ['Unique_ID', 'Category', 'Age', 'Sex', 'Buddy']
df = pd.DataFrame(columns=columns)
Sexes = ['female', 'male']
df.Sex = np.random.choice(a=Sexes, size=n, p=[0.6, 0.4])
list_Category = [1,2,3,4]
df.Category = np.random.choice(a=list_category, size=n, p=[0.3, 0.4, 0.2, 0.1])
buddy_list = ['buddy_1', 'buddy_2']
def assign_buddy(Category_prob_list):
"""
takes in a Category value
return: Buddy
"""
if df['Category'] == list_Category[0]:
df['Buddy'] = np.random.choice(a=buddy_list, size=n, p=[0.1, 0.9])
return df['Buddy']
elif df['Category'] == list_Category[1]:
df['Buddy'] = np.random.choice(a=buddy_list, size=n, p=[0.3, 0.7])
return df['Buddy']
elif df['Category'] == list_Category[2]:
df['Buddy'] = np.random.choice(a=buddy_list, size=n, p=[0.7, 0.3])
return df['Buddy']
elif df['Category'] == list_Category[3]:
df['Buddy'] = np.random.choice(a=buddy_list, size=n, p=[0.9, 0.1])
return df['Buddy']
else:
pass
# should apply assign_buddy to each row in df
df['Category'].apply((assign_buddy))
i have a dictionary of probabilities for assign_buddy, but can not figure out the map and apply logic despite all the documentation .
i've tried creating a function that returns probabilities from d to be passed to the argument p in np.random.choice, but it's not working.
# key is category label and values are probabilities for np.random.choice
d = {1: [0.1, 0.9], 2: [0.3, 0.7], 3: [0.7, 0.3], 4: [0.9, 0.1]}
any insight appreciated!
Try this
n = 20
columns = ['Unique_ID', 'Category', 'Age', 'Sex', 'Buddy']
df = pd.DataFrame(columns=columns)
list_category = [1,2,3,4]
buddy_list = ['buddy_1', 'buddy_2']
Sexes = ['female', 'male']
df.Sex = np.random.choice(a=Sexes, size=n, p=[0.6, 0.4])
df.Category = np.random.choice(list_category, size=n, p=[0.3, 0.4, 0.2, 0.1])
d = {1: [0.1, 0.9], 2: [0.3, 0.7], 3: [0.7, 0.3], 4: [0.9, 0.1]}
for val in list_category:
sz = (df["Category"] == val).sum() # find the size for array to create
# use `loc` to select places you want to replace
df.loc[df["Category"] == val,'Buddy'] = np.random.choice(
buddy_list, sz, p=d[val])