Rolling up each columnar value in a matrix

user_3792360 0
Viewed 100

Here is the data.

 day_value = {
        'android':[1,0,0,0,0,0,1],
         'iphone':[0,1,0,1,0,0,0],
            'web':[0,1,1,0,1,0,0],
     }


device_rollup = {
    'overall':['iphone','android','web'],
     'mobile':['iphone','android'],
 }

rollup_l7 = {
    'overall': 6,
     'mobile': 4,
 }

If the values are same on each column we must rollup as 1 not add it up. The total for overall should be 6 ((1,0,0)+(0,1,1)+(0,0,1)+(0,1,0)+ (0,0,1)+ (1,0,0)-> 1+1+1+1+1+1 =6)

Now I can add up all the values like this.

overall_val = sum(sum(v) for k,v in day_value.items() if k in device_rollup['overall'] )
mobile_val = sum(sum(v) for k,v in day_value.items() if k in device_rollup['mobile'] )

rollup_l7= {'overall':overall_val, 'mobile':mobile_val}
print(rollups_l7)

But, I don't want to add up instead I would like to roll up each columnar value.

Do we need to convert the list to binary and then covert back to integer?

I am not quite sure how I can achieve in regular python without using numpy

rollup pandas python
2 Answers

I think I figured out the answer.

overall = sum([x | y | z  for x,y,z in zip(day_value['android'], day_value['iphone'],day_value['web'])])
mobile = sum([x | y for x,y in zip(day_value['android'], day_value['iphone'])])

rollup_l7 = {'overall':overall, 'mobile':mobile}
print(rollup_l7)
user_37923600

Solution

I don't know typically how big the dictionary day_value will be. But you could use pandas as an alternative nonetheless. The following code block will process all the data as you explained.

# pip install pandas
import pandas as pd

## define label categories for modularity
categories = {
    'overall': ['android', 'iphone', 'web'], 
    'mobile': ['android', 'iphone'], 
    'rollup': ['overall', 'mobile']  
}

## create a dataframe with input data
## and process for 'overall' and 'mobile'
df = pd.DataFrame(day_value)
df['overall'] = df[categories['overall']].sum(axis=1) > 0
df['mobile'] = df[categories['mobile']].sum(axis=1) > 0

## evaluate totals of all columns in df
total = df.sum(axis=0).astype(int)

## update device_rollup
device_rollup = {'overall': [], 'mobile': []}
for k, v in total[categories['overall']].to_dict().items(): 
    print(k, v)
    if v > 0:
        device_rollup['overall'].append(k)
        if k in categories['mobile']:
            device_rollup['mobile'].append(k) 

## update rollup_l7
rollup_l7 = total[categories['rollup']].to_dict()

Output:

## df: dataframe
# print(df)
   android  iphone  web  overall  mobile
0        1       0    0     True    True
1        0       1    1     True    True
2        0       0    1     True   False
3        0       1    0     True    True
4        0       0    1     True   False
5        0       0    0    False   False
6        1       0    0     True    True

## total
# print(total.to_dict())
{'android': 2, 'iphone': 2, 'mobile': 4, 'overall': 6, 'web': 3}

## device_rollup
# print(device_rollup)
{
    'mobile': ['android', 'iphone'], 
    'overall': ['android', 'iphone', 'web']
}

## rollup_l7
# print(rollup_l7)
{'mobile': 4, 'overall': 6}

Data

day_value = {
    'android': [1,0,0,0,0,0,1],
    'iphone':  [0,1,0,1,0,0,0],
    'web':     [0,1,1,0,1,0,0],
}
user_84748940
Related