Python's Evolution on Stack Overflow: An Empirical Analysis of Topic Trends

This is the experiment data of the thesis "Python's Evolution on Stack Overflow: An Empirical Analysis of Topic Trends". This dataset is from the open source "Stack Exchange Data Dump", which shares the latest posts information monthly. The data downloaded for this study was published on September 5, 2023. Raw Data This is the raw data of "Stack Exchange Data Dump". We just separate them by year. The original data source is from : https://archive.org/details/stackexchange Training Data Each set contains two files, Training Set and Rater, Training Set is the content of the training set, i.e., the file in which the model is directly trained, and the Grade column is the result of the machine's prediction.The Rater file contains the data of the manual scoring, and contains the following sections: 1. the rating data for each rater, with the specific column name Rater- rater number (e.g. Rater-1) 2. the machine-predicted rating Predicted-Grade 3. the average of the manual ratings (after removing the highest and lowest) Predicted Data(xlsx) Inside is the result of each year after heat calculation, some of the column names are the original column names in the xml file, some of the column names are added for better prediction of the heat value: BodyLength: The length of post content. CreateDate: The date calculation is converted to the number of days from column 'CreationDate' to September 30, 2023 (i.e., the last day of the month in which the dataset was published) to avoid the date of some posts being converted to zero. ActiveDate: The date calculation is converted to the number of days from column 'LastActivityDate' to September 30, 2023 (i.e., the last day of the month in which the dataset was published) to avoid the date of some posts being converted to zero. Score-1: The result of dividing the total number of 'Score' by the 'CreateDate' (i.e., the value of scores earned per day) Predicted_Grade: Grades predicted for each post GradeScore: Each grade's value本数据集为论文《Python在Stack Overflow上的演进：主题趋势实证分析》的实验数据。该数据集源自开源的**Stack Exchange 数据转储（Stack Exchange Data Dump）**，该资源每月更新最新帖子信息。本研究使用的数据集发布于2023年9月5日。 ### 原始数据 本部分为“Stack Exchange 数据转储”的原始数据，仅按年份进行了拆分。原始数据源地址为：https://archive.org/details/stackexchange ### 训练数据 每组数据集包含两个文件：训练集文件与评分者文件。训练集文件为模型直接训练所用的内容，其中“Grade”列为机器预测结果。评分者文件包含人工评分数据，涵盖以下内容： 1. 各评分者的评分数据，列名格式为“Rater-评分者编号”（例如“Rater-1”）； 2. 机器预测评分“Predicted-Grade”； 3. 剔除最高分与最低分后的人工评分平均值。 ### 预测数据（XLSX格式） 该文件内为各年度热度计算结果，部分列名沿用XML文件中的原始列名，部分列名为辅助热度值预测新增的自定义列名： - “BodyLength”：帖子内容的长度； - “CreateDate”：日期计算项，将原“CreationDate”列转换为距离2023年9月30日（即数据集发布当月的最后一日）的天数，以避免部分帖子日期被转换为零值； - “ActiveDate”：日期计算项，将原“LastActivityDate”列转换为距离2023年9月30日的天数，以避免部分帖子日期被转换为零值； - “Score-1”：将“Score”总得分除以“CreateDate”得到的结果（即每日获得的得分值）； - “Predicted_Grade”：为各帖子预测的评分； - “GradeScore”：各评分对应的分值。