Skip to content

Latest commit

 

History

History
124 lines (96 loc) · 3.28 KB

creation-algorithmes.md

File metadata and controls

124 lines (96 loc) · 3.28 KB
Raw

Algorithm creation

You can create an algorithm in 2 steps:

  1. Create a script
  2. Call in it the configuration file

(A sample call from a custom function is defined in /tests folder.)

Script creation

The script created must be a simple python file, in the same directory as the configuration file.

For documentation purposes, we will call it custom_algorithm.

The structure is the following:

  • Has to contain a main function
  • The architecture has to be the following:
def main(df, options):
    ...
    ...
    return df

You can now use all pandas transformations on the dataframe. We will provide you a list of common transformations to use, but you can now explore our algorithms here.

  • options is a dictionnary of custom parameters you choose.

You can specify parameters in the configuration file. For example:

print(options['group_by'])

Can be specified in the configuration file by:

    { 
        "name": "", 
        "options": { "group_by": "latitude" }  
    }

Calling the script in the configuration file

The name has to be the script filename, in our case from the exemple above:

...
"algorithms": [
    { 
        "name": "custom_algorithm", 
        "options": { "group_by" : "latitude" }  
    }
]
...

Script example

This is an example of script and how we call it in the configuration file. We will call it example.py

def main(df, options):
    """
    Pseudo algorithm that replaces each specified row by an alias
    """
    for row in options.columns:
        df.iloc[:, row] = options.alias
    return df

Its call in thz configuration file will be the following:

...
"algorithms": [
    { 
        "name": "example", 
        "options": { "columns" : [1], "alias": "DEL" }  
    }
]
...

If we apply the script over the dataframe below:

     id                 date  latitude  longitude
0     1  2015-03-04 00:35:48  0.527129   0.358134
1     1  2015-03-04 00:35:49  0.341051   0.329298
2     1  2015-03-04 00:35:50  0.705571   0.122502
3     1  2015-03-04 00:35:52  0.618180   0.567444
4     1  2015-03-04 00:35:54  0.686230   0.560744
..   ..                  ...       ...        ...
494   1  2015-03-04 01:03:28  0.788060   0.225917
495   1  2015-03-04 01:03:30  0.668119   0.718251
496   1  2015-03-04 01:03:32  0.173801   0.770294
497   1  2015-03-04 01:03:33  0.835937   0.000620
498   1  2015-03-04 01:03:35  0.350427   0.936959

The resulting dataframe will be:

     id                 date  latitude  longitude
0     DEL  2015-03-04 00:35:48  0.527129   0.358134
1     DEL  2015-03-04 00:35:49  0.341051   0.329298
2     DEL  2015-03-04 00:35:50  0.705571   0.122502
3     DEL  2015-03-04 00:35:52  0.618180   0.567444
4     DEL  2015-03-04 00:35:54  0.686230   0.560744
..   ..                  ...       ...        ...
494   DEL  2015-03-04 01:03:28  0.788060   0.225917
495   DEL  2015-03-04 01:03:30  0.668119   0.718251
496   DEL  2015-03-04 01:03:32  0.173801   0.770294
497   DEL  2015-03-04 01:03:33  0.835937   0.000620
498   DEL  2015-03-04 01:03:35  0.350427   0.936959