根据日期特征预测概率 [英] Predict the probability based on date feature

问题描述

你好团队

目前，我正在使用两类Logistic回归算法"来简单地检查订单是否被接受.

Currently I am using Two Class Logistic Regression Algorithm to simply check the order will get accept or not. 

但是随着需求的变化，我现在尝试获取特定订单的概率，该概率将在指定日期以后被接受或不被接受.我应该如何准备数据，以便与其他所有产品参数一起传递未来的日期 模型会相应地回应我.(未来概率)

But as requirement change bit now I am trying to get the probability of particular order, which will be accepted or not in future for provided date. How should I prepare the data so that along with all other product parameters I can pass future date as well and model will respond me accordingly.(Future probability)

在我目前的数据集中，我没有任何日期列，那么该如何处理呢?还是需要更改方法或模型?

In my present dataset I don't have any date column so how can I handle this or do I need to change the approach or model?

2018年12月25日 日期?

For ex. If user send product related data like product id, product name, product price, product discount, salesperson, state etc. along with future date like 25Dec2018. Then if the current probability (31Oct2018) for this request is 0.75% then what would be on this  25Dec2018 date? 

在圣诞节期间的预期概率为0.88％.

请提供您的想法.我应该选择哪种徽标，因为我不想像需求预测中那样预测价值.

谢谢！

问候

RPBH

推荐答案

RPBH，

通过添加日期"来提高预测的准确性.维度，您需要在日期或日期中添加一列用于日期或与特殊日期(例如节日/场合)相对应的内容.

这将是相关的，因为某些产品在某些节日/场合可能会卖出更多，而在其他节日/场合下卖得会少.因此，您的训练数据的结构可能与此类似:

产品编号|价格|折扣|营业员|州|场合/节日

Hi RPBH,

To increase accuracy of your prediction by adding the "date" dimension to it, you need to have a column for either date or something corresponding to special dates like festivals/occasions in your training data.

This would be relevant as some product may sell more on some festival/occasion and less on some other. So your training data could be something similar in structure to this :

Product Id | price | discount | salesperson | state | occasion/festival

这篇关于根据日期特征预测概率的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！