Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,425 @@
|
|
1 |
import streamlit as st
|
|
|
|
|
2 |
|
3 |
-
|
4 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import streamlit as st
|
2 |
+
import numpy as np
|
3 |
+
import plotly.figure_factory as ff
|
4 |
|
5 |
+
import plotly.graph_objects as go
|
6 |
+
import plotly.express as px
|
7 |
+
|
8 |
+
|
9 |
+
import requests
|
10 |
+
import json
|
11 |
+
import pandas as pd
|
12 |
+
import shutil
|
13 |
+
import os
|
14 |
+
from openai import AzureOpenAI
|
15 |
+
import base64
|
16 |
+
|
17 |
+
|
18 |
+
# st.page_link("report.py", label="Home", icon="🏠")
|
19 |
+
# st.page_link("pages/page_1.py", label="Page 1", icon="1️⃣")
|
20 |
+
# st.page_link("pages/page_2.py", label="Page 2", icon="2️⃣", disabled=True)
|
21 |
+
|
22 |
+
ACCOUNT_ID = "act_416207949073936"
|
23 |
+
PAGE_ID = "63257509478"
|
24 |
+
OPENAI_API = os.getenv("OPENAI_API")
|
25 |
+
ACCESS_TOKEN = os.getenv("ACCESS_TOKEN")
|
26 |
+
BIG_DATASET = None
|
27 |
+
|
28 |
+
ANALYSIS_TYPE = {
|
29 |
+
"OUTCOME_SALES": "ROAS",
|
30 |
+
}
|
31 |
+
|
32 |
+
API_BASE = 'https://bestever-vision.openai.azure.com/'
|
33 |
+
DEPLOYMENT_NAME = 'vision'
|
34 |
+
API_VERSION = '2023-12-01-preview' # this might change in the future
|
35 |
+
API_URL = f"{API_BASE}openai/deployments/{DEPLOYMENT_NAME}/extensions"
|
36 |
+
|
37 |
+
client = AzureOpenAI(
|
38 |
+
api_key=OPENAI_API,
|
39 |
+
api_version=API_VERSION,
|
40 |
+
base_url=API_URL,
|
41 |
+
)
|
42 |
+
|
43 |
+
def encode_image(image_path):
|
44 |
+
with open(image_path, "rb") as image_file:
|
45 |
+
return base64.b64encode(image_file.read()).decode('utf-8')
|
46 |
+
|
47 |
+
|
48 |
+
def call_gpt_vision(client, images_path, user_prompt):
|
49 |
+
"""Call the GPT4 Vision API to generate tags."""
|
50 |
+
images_content = [
|
51 |
+
{
|
52 |
+
"type": "image_url",
|
53 |
+
"image_url": {
|
54 |
+
"url": f"data:image/jpeg;base64,{encode_image(image_path)}",
|
55 |
+
},
|
56 |
+
}
|
57 |
+
for image_path in images_path
|
58 |
+
]
|
59 |
+
user_content = [
|
60 |
+
{"type": "text", "text": user_prompt},
|
61 |
+
]
|
62 |
+
user_content += images_content
|
63 |
+
response = client.chat.completions.create(
|
64 |
+
model=DEPLOYMENT_NAME,
|
65 |
+
messages=[
|
66 |
+
{"role": "user", "content": user_content},
|
67 |
+
],
|
68 |
+
max_tokens=2000,
|
69 |
+
)
|
70 |
+
return response
|
71 |
+
|
72 |
+
|
73 |
+
def parse_tags_from_content(response):
|
74 |
+
"""Parse the tags from the response."""
|
75 |
+
tags = []
|
76 |
+
content = response.choices[0].message.content
|
77 |
+
for full_tag in content.split("\n"):
|
78 |
+
splitted_fields = full_tag.split(":")
|
79 |
+
if len(splitted_fields) < 2:
|
80 |
+
continue
|
81 |
+
tag_name = splitted_fields[0]
|
82 |
+
tag_details = ":".join(splitted_fields[1:])
|
83 |
+
tag_element = {"name": tag_name, "metadata": {"details": tag_details}}
|
84 |
+
tags.append(tag_element)
|
85 |
+
return tags
|
86 |
+
|
87 |
+
|
88 |
+
def get_campaigns(account_id):
|
89 |
+
url = f"{account_id}/insights"
|
90 |
+
params = {
|
91 |
+
"date_preset": "last_90d",
|
92 |
+
"fields": "campaign_id,campaign_name,impressions,spend,objective",
|
93 |
+
"level": "campaign",
|
94 |
+
"access_token": ACCESS_TOKEN,
|
95 |
+
}
|
96 |
+
return call_graph_api(url, params)
|
97 |
+
|
98 |
+
|
99 |
+
def get_adsets(campaign_id):
|
100 |
+
url = f"{campaign_id}/insights"
|
101 |
+
params = {
|
102 |
+
"date_preset": "last_90d",
|
103 |
+
"fields": "adset_id,adset_name,impressions,spend",
|
104 |
+
"level": "adset",
|
105 |
+
"access_token": ACCESS_TOKEN,
|
106 |
+
}
|
107 |
+
return call_graph_api(url, params)
|
108 |
+
|
109 |
+
|
110 |
+
def get_ads(adset_id):
|
111 |
+
url = f"{adset_id}/insights"
|
112 |
+
params = {
|
113 |
+
"date_preset": "last_90d",
|
114 |
+
"fields": "ad_name,ad_id,impressions,spend,video_play_actions,video_p25_watched_actions,video_p50_watched_actions,video_p75_watched_actions,video_p100_watched_actions,video_play_curve_actions,purchase_roas",
|
115 |
+
"breakdowns": "age,gender",
|
116 |
+
"limit": 1000,
|
117 |
+
"level": "ad",
|
118 |
+
"access_token": ACCESS_TOKEN,
|
119 |
+
}
|
120 |
+
return call_graph_api(url, params)
|
121 |
+
|
122 |
+
|
123 |
+
def save_image_from_url(url, filename):
|
124 |
+
res = requests.get(url, stream = True)
|
125 |
+
|
126 |
+
if res.status_code == 200:
|
127 |
+
with open(filename,'wb') as f:
|
128 |
+
shutil.copyfileobj(res.raw, f)
|
129 |
+
return True
|
130 |
+
return False
|
131 |
+
|
132 |
+
def get_creative_assets(ad_id):
|
133 |
+
# checking if the asset already exists
|
134 |
+
if os.path.exists(f'assets/{ad_id}.png') or os.path.exists(f'assets/{ad_id}.mp4') or os.path.exists(f'assets/{ad_id}.jpg'):
|
135 |
+
return
|
136 |
+
url = f"{ad_id}"
|
137 |
+
params = {
|
138 |
+
"fields": "creative{video_id,id,effective_object_story_id,image_url}",
|
139 |
+
"access_token": ACCESS_TOKEN,
|
140 |
+
}
|
141 |
+
creative = call_graph_api(url, params)["creative"]
|
142 |
+
saved = False
|
143 |
+
print("-" * 10)
|
144 |
+
if "video_id" in creative:
|
145 |
+
# download video
|
146 |
+
video_id = creative["video_id"]
|
147 |
+
video_url = f"{video_id}"
|
148 |
+
video_params = {
|
149 |
+
"fields": "source",
|
150 |
+
"access_token": ACCESS_TOKEN,
|
151 |
+
}
|
152 |
+
video_source = call_graph_api(video_url, video_params)["source"]
|
153 |
+
ext = video_source.split("?")[0].split(".")[-1]
|
154 |
+
if len(ext) > 4:
|
155 |
+
ext = "mp4"
|
156 |
+
saved = save_image_from_url(video_source, os.path.join("assets", f'{ad_id}.{ext}'))
|
157 |
+
|
158 |
+
elif "image_url" in creative:
|
159 |
+
image_url = creative["image_url"]
|
160 |
+
ext = image_url.split("?")[0].split(".")[-1]
|
161 |
+
if len(ext) > 4:
|
162 |
+
ext = "png"
|
163 |
+
saved = save_image_from_url(image_url, os.path.join("assets", f'{ad_id}.{ext}'))
|
164 |
+
|
165 |
+
elif "effective_object_story_id" in creative:
|
166 |
+
object_story_url = creative["effective_object_story_id"]
|
167 |
+
object_story_params = {
|
168 |
+
"fields": "attachments",
|
169 |
+
"access_token": ACCESS_TOKEN,
|
170 |
+
}
|
171 |
+
attachments = call_graph_api(object_story_url, object_story_params)["attachments"]
|
172 |
+
if "media" in attachments:
|
173 |
+
media = attachments["media"]
|
174 |
+
if "source" in media or "video" in media:
|
175 |
+
video_url = media["video"]["source"]
|
176 |
+
ext = video_url.split("?")[0].split(".")[-1]
|
177 |
+
if len(ext) > 4:
|
178 |
+
ext = "png"
|
179 |
+
saved = save_image_from_url(video_url, os.path.join("assets", f'{ad_id}.{ext}'))
|
180 |
+
elif "image" in media:
|
181 |
+
image_url = media["image"]["src"]
|
182 |
+
ext = image_url.split("?")[0].split(".")[-1]
|
183 |
+
if len(ext) > 4:
|
184 |
+
ext = "mp4"
|
185 |
+
saved = save_image_from_url(image_url, os.path.join("assets", f'{ad_id}.{ext}'))
|
186 |
+
|
187 |
+
if not saved:
|
188 |
+
creative_url = f'{creative["id"]}'
|
189 |
+
creative_params = {
|
190 |
+
"fields": "thumbnail_url",
|
191 |
+
"access_token": ACCESS_TOKEN,
|
192 |
+
"thumbnail_width": 512,
|
193 |
+
"thumbnail_height": 512,
|
194 |
+
}
|
195 |
+
thumbnail_url = call_graph_api(creative_url, creative_params)["thumbnail_url"]
|
196 |
+
ext = thumbnail_url.split("?")[0].split(".")[-1]
|
197 |
+
if len(ext) > 4:
|
198 |
+
ext = "jpg"
|
199 |
+
saved = save_image_from_url(thumbnail_url, os.path.join("assets", f'{ad_id}.{ext}'))
|
200 |
+
|
201 |
+
def call_graph_api(url, params):
|
202 |
+
base_url = "https://graph.facebook.com/v19.0/"
|
203 |
+
response = requests.get(base_url + url, params=params)
|
204 |
+
return json.loads(response.text)
|
205 |
+
|
206 |
+
|
207 |
+
def top_n_ads(df, n=5):
|
208 |
+
ad_ids = df.head(n)["ad_id"].values
|
209 |
+
image_paths = []
|
210 |
+
for ad_id in ad_ids:
|
211 |
+
if os.path.exists(f'assets/{ad_id}.png'):
|
212 |
+
image_paths.append(f'assets/{ad_id}.png')
|
213 |
+
elif os.path.exists(f'assets/{ad_id}.mp4'):
|
214 |
+
image_paths.append(f'assets/{ad_id}.mp4')
|
215 |
+
elif os.path.exists(f'assets/{ad_id}.jpg'):
|
216 |
+
image_paths.append(f'assets/{ad_id}.jpg')
|
217 |
+
return image_paths
|
218 |
+
|
219 |
+
|
220 |
+
def perform_analysis(df, objective):
|
221 |
+
# - TS to CTR ratio analysis
|
222 |
+
# - ROAS analysis (I will see the better metric here to use)
|
223 |
+
# - Video drop off analysis
|
224 |
+
if ANALYSIS_TYPE[objective] == "ROAS":
|
225 |
+
# 3 analysis:
|
226 |
+
# general
|
227 |
+
# male
|
228 |
+
# female
|
229 |
+
|
230 |
+
df_general = df.groupby(["ad_id"]).sum()
|
231 |
+
df_general = df_general.reset_index()
|
232 |
+
df_general["relative_roas"] = df_general["purchase_roas"] / df_general["spend"]
|
233 |
+
df_general = df_general.sort_values("relative_roas", ascending=False)
|
234 |
+
|
235 |
+
image_paths = top_n_ads(df_general)
|
236 |
+
response = call_gpt_vision(client, image_paths, "You are a marketing analyst and your task is to find common features between the most performatives ads of the company. You are given the top 5 most perfomative ads, and we expect you to return 5 keywords and its explanation that defines what makes a good ad that show an excellent ROAS. Return it as a list of 5 concepts and its explanation, using the provided ads as example. Try to use nice categories to describe the features (you can use some names like `minimalist design`, `Clear message`, etc). Also, pay attention if the ads are mostly images or videos, this is important to say. The output MUST contain one concept per line. For each like, follow the structure: <concept>:<explanation>.")
|
237 |
+
image_winner_concepts = parse_tags_from_content(response)
|
238 |
+
|
239 |
+
response = call_gpt_vision(client, [], f"Following, you have the key features that makes an ad a performative ad. Your task is to group this information and summarize in a nice paragraph that will be presented to the marketing team. Be concise. Features:\n{image_winner_concepts}")
|
240 |
+
insights = response.choices[0].message.content
|
241 |
+
|
242 |
+
general_output = {"keywords": [concept["name"] for concept in image_winner_concepts], "insights": insights}
|
243 |
+
|
244 |
+
# Groupby ad_id and gender
|
245 |
+
df_male = df[df["gender"] == "male"].groupby(["ad_id"]).sum()
|
246 |
+
df_male = df_male.reset_index()
|
247 |
+
df_male["relative_roas"] = df_male["purchase_roas"] / df_male["spend"]
|
248 |
+
df_male = df_male.sort_values("relative_roas", ascending=False)
|
249 |
+
|
250 |
+
image_paths = top_n_ads(df_male)
|
251 |
+
response = call_gpt_vision(client, image_paths, "You are a marketing analyst and your task is to find common features between the most performatives ads published to men. You are given the top 5 most perfomative ads, and we expect you to return 5 keywords and its explanation that defines what makes a good ad that show an excellent ROAS. Return it as a list of 5 concepts and its explanation, using the provided ads as example. Try to use nice categories to describe the features (you can use some names like `minimalist design`, `Clear message`, etc). Also, pay attention if the ads are mostly images or videos, this is important to say. The output MUST contain one concept per line. For each like, follow the structure: <concept>:<explanation>.")
|
252 |
+
image_winner_concepts = parse_tags_from_content(response)
|
253 |
+
|
254 |
+
response = call_gpt_vision(client, [], f"Following, you have the key features that makes an ad a performative ad. Your task is to group this information and summarize in a nice paragraph that will be presented to the marketing team. Be concise. Features:\n{image_winner_concepts}")
|
255 |
+
insights = response.choices[0].message.content
|
256 |
+
|
257 |
+
male_output = {"keywords": [concept["name"] for concept in image_winner_concepts], "insights": insights}
|
258 |
+
|
259 |
+
|
260 |
+
df_female = df[df["gender"] == "female"].groupby(["ad_id"]).sum()
|
261 |
+
df_female = df_female.reset_index()
|
262 |
+
df_female["relative_roas"] = df_female["purchase_roas"] / df_female["spend"]
|
263 |
+
df_female = df_female.sort_values("relative_roas", ascending=False)
|
264 |
+
|
265 |
+
image_paths = top_n_ads(df_female)
|
266 |
+
response = call_gpt_vision(client, image_paths, "You are a marketing analyst and your task is to find common features between the most performatives ads published to women. You are given the top 5 most perfomative ads, and we expect you to return 5 keywords and its explanation that defines what makes a good ad that show an excellent ROAS. Return it as a list of 5 concepts and its explanation, using the provided ads as example. Try to use nice categories to describe the features (you can use some names like `minimalist design`, `Clear message`, etc). Also, pay attention if the ads are mostly images or videos, this is important to say. The output MUST contain one concept per line. For each like, follow the structure: <concept>:<explanation>.")
|
267 |
+
image_winner_concepts = parse_tags_from_content(response)
|
268 |
+
|
269 |
+
response = call_gpt_vision(client, [], f"Following, you have the key features that makes an ad a performative ad. Your task is to group this information and summarize in a nice paragraph that will be presented to the marketing team. Be concise. Features:\n{image_winner_concepts}")
|
270 |
+
insights = response.choices[0].message.content
|
271 |
+
female_output = {"keywords": [concept["name"] for concept in image_winner_concepts], "insights": insights}
|
272 |
+
|
273 |
+
return {
|
274 |
+
"General": general_output,
|
275 |
+
"Male": male_output,
|
276 |
+
"Female": female_output,
|
277 |
+
}
|
278 |
+
|
279 |
+
def format_adsets(campaign_id):
|
280 |
+
st_campaigns.empty()
|
281 |
+
adsets = get_adsets(campaign_id)
|
282 |
+
with st_adsets.container():
|
283 |
+
st.title("Adsets")
|
284 |
+
for adset in adsets["data"]:
|
285 |
+
with st.popover(adset["adset_name"]):
|
286 |
+
st.markdown("**Impressions**: " + str(adset["impressions"]))
|
287 |
+
st.markdown("**Total Spend**: US$" + str(adset["spend"]))
|
288 |
+
st.button(
|
289 |
+
"View Ads",
|
290 |
+
key=adset["adset_name"],
|
291 |
+
on_click=format_ads,
|
292 |
+
kwargs={"adset_id": adset["adset_id"]},
|
293 |
+
)
|
294 |
+
|
295 |
+
|
296 |
+
def format_ads(adset_id):
|
297 |
+
st_adsets.empty()
|
298 |
+
BIG_DATASET = None
|
299 |
+
ads = get_ads(adset_id)
|
300 |
+
df_ads = pd.DataFrame(ads["data"])
|
301 |
+
options = ["gender"] #st.multiselect(
|
302 |
+
# "Which breakdowns do you want to see?", ["gender", "age"], ["gender"]
|
303 |
+
# )
|
304 |
+
df_ads["spend"] = df_ads["spend"].astype(float)
|
305 |
+
df_ads["impressions"] = df_ads["impressions"].astype(float)
|
306 |
+
video_cols = ["video_play_actions","video_p25_watched_actions","video_p50_watched_actions","video_p75_watched_actions","video_p100_watched_actions"]
|
307 |
+
for col in video_cols:
|
308 |
+
if col in df_ads.columns:
|
309 |
+
df_ads[col] = df_ads[col].apply(lambda x: float(x[0].get("value", 0)) if isinstance(x, list) else 0)
|
310 |
+
|
311 |
+
if "purchase_roas" in df_ads.columns:
|
312 |
+
df_ads["purchase_roas"] = df_ads["purchase_roas"].apply(lambda x: float(x[0].get("value", 0)) if isinstance(x, list) else 0)
|
313 |
+
|
314 |
+
if BIG_DATASET is None:
|
315 |
+
BIG_DATASET = df_ads
|
316 |
+
else:
|
317 |
+
BIG_DATASET = pd.concat([BIG_DATASET, df_ads])
|
318 |
+
BIG_DATASET.to_csv("big_dataset.csv")
|
319 |
+
with st_ads.container():
|
320 |
+
with st.expander("See analysis", expanded=False):
|
321 |
+
analysis = st.empty()
|
322 |
+
|
323 |
+
for i, ad in enumerate(df_ads["ad_id"].unique()):
|
324 |
+
get_creative_assets(ad)
|
325 |
+
ad_name = df_ads[df_ads["ad_id"] == ad]["ad_name"].values[0]
|
326 |
+
with st.popover(ad_name):
|
327 |
+
tab1, tab2, tab3 = st.tabs(["Creative", "Analytics", "Video Analysis"])
|
328 |
+
df_tmp = df_ads[df_ads["ad_id"] == ad]
|
329 |
+
with tab2:
|
330 |
+
if len(options) >= 1:
|
331 |
+
label = ["Total impressions"]
|
332 |
+
source = []
|
333 |
+
target = []
|
334 |
+
value = []
|
335 |
+
for option in options:
|
336 |
+
df_g_tmp = df_tmp.groupby(option).sum()
|
337 |
+
df_g_tmp = df_g_tmp.reset_index()
|
338 |
+
for imp, v in df_g_tmp[["impressions", option]].values:
|
339 |
+
label.append(v)
|
340 |
+
source.append(0)
|
341 |
+
target.append(len(label) - 1)
|
342 |
+
value.append(imp)
|
343 |
+
|
344 |
+
fig = go.Figure(
|
345 |
+
data=[
|
346 |
+
go.Sankey(
|
347 |
+
node=dict(
|
348 |
+
pad=15,
|
349 |
+
thickness=20,
|
350 |
+
line=dict(color="black", width=0.5),
|
351 |
+
label=label,
|
352 |
+
color="blue",
|
353 |
+
),
|
354 |
+
link=dict(
|
355 |
+
source=source, target=target, value=value
|
356 |
+
),
|
357 |
+
)
|
358 |
+
]
|
359 |
+
)
|
360 |
+
fig.update_layout(title_text="Basic Sankey Diagram", font_size=10)
|
361 |
+
st.plotly_chart(fig, use_container_width=True)
|
362 |
+
|
363 |
+
if "purchase_roas" in df_tmp.columns:
|
364 |
+
df_roas = df_tmp.groupby(options)[["spend","purchase_roas"]].sum().reset_index().sort_values("purchase_roas", ascending=False)
|
365 |
+
print(df_roas)
|
366 |
+
values = [str(v) for v in df_tmp[options].values]
|
367 |
+
fig = go.Figure(data=[
|
368 |
+
go.Bar(name='ROAS', x=values, y=df_roas["purchase_roas"]),
|
369 |
+
go.Bar(name='Spend', x=values, y=df_roas["spend"])
|
370 |
+
])
|
371 |
+
# Change the bar mode
|
372 |
+
fig.update_layout(barmode='group')
|
373 |
+
st.plotly_chart(fig, use_container_width=True)
|
374 |
+
|
375 |
+
with tab3:
|
376 |
+
if "video_play_actions" in df_tmp.columns:
|
377 |
+
values = df_ads[["ad_id","video_play_actions","video_p25_watched_actions","video_p50_watched_actions","video_p75_watched_actions","video_p100_watched_actions"]].groupby("ad_id").get_group(ad).sum().values[1:]
|
378 |
+
labels = ["Total video plays","Video plays until 25%","Video plays until 50%","Video plays until 75%","Video plays until 100%"]
|
379 |
+
print(values)
|
380 |
+
if values[0] > 0:
|
381 |
+
st.plotly_chart(create_video_plays_funnel(values, labels), use_container_width=True)
|
382 |
+
with tab1:
|
383 |
+
if os.path.exists(f'assets/{ad}.png'):
|
384 |
+
st.image(f'assets/{ad}.png', caption='Creative', use_column_width=True)
|
385 |
+
elif os.path.exists(f'assets/{ad}.mp4'):
|
386 |
+
st.video(f'assets/{ad}.mp4')
|
387 |
+
elif os.path.exists(f'assets/{ad}.jpg'):
|
388 |
+
st.image(f'assets/{ad}.jpg', caption='Creative', use_column_width=True)
|
389 |
+
|
390 |
+
with analysis.container():
|
391 |
+
report = perform_analysis(df_tmp, "OUTCOME_SALES")
|
392 |
+
tabs = st.tabs(report.keys())
|
393 |
+
tabs_names = list(report.keys())
|
394 |
+
for i, tab in enumerate(tabs):
|
395 |
+
with tab:
|
396 |
+
st.multiselect("", report[tabs_names[i]]["keywords"], report[tabs_names[i]]["keywords"], key=f"{ad}_{i}")
|
397 |
+
st.write(report[tabs_names[i]]["insights"])
|
398 |
+
|
399 |
+
def create_video_plays_funnel(funnel_data, funnel_title):
|
400 |
+
fig = go.Figure(go.Funnel(
|
401 |
+
y = funnel_title,
|
402 |
+
x = funnel_data))
|
403 |
+
return fig
|
404 |
+
|
405 |
+
if "initiated" not in st.session_state:
|
406 |
+
st.session_state["initiated"] = False
|
407 |
+
|
408 |
+
if not st.session_state["initiated"]:
|
409 |
+
st_campaigns = st.empty()
|
410 |
+
st_adsets = st.empty()
|
411 |
+
st_ads = st.empty()
|
412 |
+
st.session_state["initiated"] = True
|
413 |
+
with st_campaigns.container():
|
414 |
+
st.title("Campaigns")
|
415 |
+
for c in (get_campaigns(ACCOUNT_ID))["data"]:
|
416 |
+
with st.popover(c["campaign_name"]):
|
417 |
+
st.markdown("**Impressions**: " + str(c["impressions"]))
|
418 |
+
st.markdown("**Total Spend**: US$" + str(c["spend"]))
|
419 |
+
st.markdown("**Objective**: " + str(c["objective"]))
|
420 |
+
st.button(
|
421 |
+
"View Adsets",
|
422 |
+
key=c["campaign_name"],
|
423 |
+
on_click=format_adsets,
|
424 |
+
kwargs={"campaign_id": c["campaign_id"]},
|
425 |
+
)
|