1. The pacing algorithm is located in the folder Pacing_project

2. All steps to construct the algorithm are notebooks situated in the folder Pacing_notebooks

3. The Report folder contains the internship report and slides

4. The Miscellaneous folder contains miscellaneous things such as work on auction theory

This is the main folder of the project. It contains all useful scripts about the pacing algorithm.
pacing_class_tz.py is the script that contains the final algorithm.
external_functions_tz.pyis the script that contains functions used to simulate the algorithm.
execution_tz.py allows to simulate the algorithm on a dataframe (give any dataframe on the data folder).
api_rest.py is the script to generate the API of the algorithm. It allows to launch a local server.
exec_api.py is the script that simulates the API (with a dataframe of br situated in the data folder).
test_basics.py is basic unit tests on the API.

Note: pacing_class.py, external_functions.py and execution.py are the same that those with the suffix tz, but the algorithm doesn't allow multiple time zones.

To start a local server in order to use the API, launch on a terminal the following command:

python api_rest.py

POST method

1. Initialise a campaign:

curl --request POST \
  --url http://127.0.0.1:8000/campaign \
  --header 'content-type: application/json' \
  --data '{
	"cpid": String ID 
}'

2. Initialise a line item:

curl --request POST \
  --url http://127.0.0.1:8000/campaign/1/init \
  --header 'content-type: application/json' \
  --data '{
	"budget": Budget,
	"start": String date,
	"end": String date,
	"liid": String ID
}'

3. Send a bid request

curl --request POST \
  --url http://127.0.0.1:8000/li/1/br \
  --header 'content-type: application/json' \
  --data '{
	"tz": String time zone,
	"brid": ID of BR,
	"imps": Number of impressions,
	"cpm": CPM
}'

4. Send notification

curl --request POST \
  --url http://127.0.0.1:8000/li/1/notif \
  --header 'content-type: application/json' \
  --data '{
	"status": Status: must be 'win' or 'lose',
	"brid": ID of BR
}'

GET method
Note that all output are in JSON format.

1. Get campaigns list

curl --request GET \
  --url http://127.0.0.1:8000/campaign

For example the if there is a campaign with the ID "1" and 3 line items for this campaign:

{
  "campaigns": {
    "1": 3
  }
}

2. Get line items list

curl --request GET \
  --url http://127.0.0.1:8000/li

The output for 3 line items would be the following:

{
  "status": "ok",
  "LineItems": [
    "1",
    "2",
    "3"
  ]
}

3. Get general status

curl --request GET \
  --url http://127.0.0.1:8000/li/1/status

The general status of a line item returns the sum of expenditures and the total remaining budget:

{
  "spent": 0,
  "remaining": 10000
}

4. Get status detailed by time zone

curl --request GET \
  --url http://127.0.0.1:8000/li/1/status/tz

The output is the expenditure per time zone. Assuming that we have spent $1 in the time zone America/New_York, the output would be:

{
  "spent": {
    "Europe/Paris": 0,
    "America/New_york": 1.0
  },
  "remaining": {
    "Europe/Paris": 5000.0,
    "America/New_york": 4999.0
  }
}

5. Get status of a precised time zone

curl --request GET \
  --url http://127.0.0.1:8000/li/1/status/tz/STRING_TIMEZONE

Note: The / of the timezone in the URL should be replaced by --. For example if you want the time zone America/New_York:

curl --request GET \
  --url http://127.0.0.1:8000/li/1/status/tz/America--New_York

The output would be then:

{
  "spent": 1.0,
  "remaining": 4999.0
}

Test the API
You can launch unit tests on the API by starting the local server and then execute the following command:

py.test .

To test the API in "real" conditions, you can start the local server, and if you have a br dataframe, just launch the exec_api.py script.

python exec_api.py