|
所在平台: Coursera |
课程主页: https://www.coursera.org/learn/ml-safety-stock
课程评论：没有评论
课程名称：最终项目：预测安全库存 课程概述：在本课程中，我们将对产品使用情况进行预测，并计算最佳安全库存储量。课程开始时，我们将分析来自三个不同大陆多家商店的鞋类销售时间序列数据。我们将通过进行分组和比较每家商店内的产品，寻找数据中独特的见解和有趣的信息。然后，我们将使用季节性自回归积分滑动平均（SARIMA）模型对未来销售进行预测。除了进行预测外，我们还将分析提供的统计数据（如p值），以判断使用SARIMA模型进行预测的可行性。随后，我们将调整模型的超参数，以获得更好的结果和更高的统计显著性。最后，我们将根据数据进行每月使用量的预测，并基于涉及交货时间的公式计算安全库存。 课程大纲： 第一部分：使用Pandas和Groupby进行探索性数据分析 描述：在本模块中，我们将通过探索明显的分组并识别产品变化，来熟悉我们的数据集。我们将发现哪些产品在何处销售，并为使用时间序列预测和安全库存预测做好准备。 第二部分：使用SARIMA进行需求预测 描述：在本模块中，我们将使用SARIMA模型对未来销售进行预测。然后，我们将可视化一些预测的销售数据，并评估所选模型的准确性和可行性。 第三部分：计算安全库存 描述：在本模块中，我们将通过每月使用量和交货时间来计算安全库存。我们将通过对产品进行分组来找到更准确的使用量。最后，我们将结合已知公式和数据见解来计算每种产品的安全库存。
Part: 1
Title:Exploratory Data Analysis Using Pandas and Groupby
Description:In this module, we'll get acquainted with our dataset by exploring some of the most obvious groupings and identifying the variation in products. We'll discover which products sell where and prepare ourselves to use timeseries forecasts and safety stock predictions.
Part: 2
Title:Demand Predictions Using SARIMA
Description:In this module, we'll use the SARIMA model to make predictions on future sales. We'll then visualize some of these predicted sales before evaluating the accuracy and viability of our chosen model.
Part: 3
Title:Calculating Safety Stock
Description:In this module, we'll finish the project by calculating safety stock from monthly usage and lead times. We'll start by grouping products in order to find more accurate usage numbers. Then, we'll conclude by using the known formula along with our insights from the data in the calculation of safety stock for each product.
In this course, we'll make predictions on product usage and calculate optimal safety stock storage. We'll start with a time series of shoe sales across multiple stores on three different continents. To begin, we'll look for unique insights and other interesting things we can find in the data by performing groupings and comparing products within each store. Then, we'll use a seasonal autoregressive integrated moving average (SARIMA) model to make predictions on future sales. In addition to making predictions, we'll analyze the provided statistics (such as p-score) to judge the viability of using the SARIMA model to make predictions. Then, we'll tune the hyper-parameters of the model to garner better results and higher statistical significance. Finally, we'll make predictions on safety stock by looking to the data for monthly usage predictions and calculating safety stock from the formula involving lead times.
