代码拉取完成，页面将自动刷新
Reasonp is a data analysis tool for finding reasons for changes in data. You can use it to explore the reasons for the decline in metrics, and you can also use it to analyze the differences in metrics between different objects. With a simple operation, it can automatically search for potential causes and find inspiration for your analysis.
You need to provide a fine-grained data so that you can find the reasons for you through segmentation. For example, to analyze the fluctuation of traffic, data of various fields of each visit is required.
Example:
pip install reasonp
import reasonp as rp
traffic_data = [
("A", "P1", "2021-03-02", 1),
("B", "P1", "2021-03-02", 1),
("A", "P2", "2021-03-02", 1),
("B", "P2", "2021-03-02", 1),
("A", "P1", "2021-03-03", 3),
("B", "P1", "2021-03-03", 4),
("A", "P2", "2021-03-03", 1),
("B", "P2", "2021-03-03", 2),
("C", "P2", "2021-03-03", 1),
]
df = pd.DataFrame(traffic_data, columns=["source", "platform", "date", "orders"])
# You want to find the reason for the change in traffic between March 3rd and March 2nd
rdf = rp.find_reason(
df,
compare_column="date",
compare_values=["2021-03-03","2021-03-02"],
compare_metric=("orders", "sum"),
)
print(rdf)
Check the result
| search_column | target_column_value | compare_column_value | target_metric_value | compare_metric_value | metric_change | reason_coef |
|---|---|---|---|---|---|---|
| platform | P1 | P1 | 7 | 2 | 5 | 0.714286 |
| source | B | B | 6 | 2 | 4 | 0.571429 |
| source | A | A | 4 | 2 | 2 | 0.285714 |
| platform | P2 | P2 | 4 | 2 | 2 | 0.285714 |
| source | C | nan | 1 | 0 | 1 | 0.142857 |
The traffic change of platform P1 may be the main reason.Because it had 5 more traffic on March 3rd than on March 2nd.
$$
reason_coef = metric_change / total_change
$$
Build
python -m build
twine upload dist/*
此处可能存在不合适展示的内容，页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容，可点击提交进行申诉，我们将尽快为您处理。