Duyu commited on
Commit
954d86f
·
verified ·
1 Parent(s): aa7d455

Upload 3 README files

Browse files
Files changed (3) hide show
  1. README.md +37 -0
  2. README.vi.md +33 -0
  3. README.zh-CN.md +32 -0
README.md ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ task_categories:
3
+ - time-series-forecasting
4
+ ---
5
+
6
+ # Time Series Forecasting Benchmark Datasets
7
+
8
+ ## Documentation Language
9
+
10
+ [**简体中文**](./README.zh-CN.md) | [**English**](./README.md) | [**Tiếng Việt**](./README.vi.md)
11
+
12
+ ## Dataset Download
13
+
14
+ * [**https://huggingface.co/datasets/Duyu/Time-Series-Forecasting-Benchmark-Datasets/tree/main**](https://huggingface.co/datasets/Duyu/Time-Series-Forecasting-Benchmark-Datasets/tree/main)
15
+
16
+ * [**https://github.com/duyu09/TimeSeries-Forecasting-Dataset/releases/download/v1.0.0/dataset.7z**](https://github.com/duyu09/TimeSeries-Forecasting-Dataset/releases/download/v1.0.0/dataset.7z)
17
+
18
+ ## Dataset Description
19
+
20
+ * **`ETT`** The Electricity Transformer Temperature (ETT) dataset serves as a critical benchmark for evaluating electric power forecasting. It comprises two years of data collected from two separate counties in China. To analyze the impact of temporal granularity, the dataset is divided into four subsets with different sampling frequencies: ETTh1 and ETTh2 are sampled at 1-hour intervals, while ETTm1 and ETTm2 are sampled at 15-minute intervals. Each data point contains six power load-related features along with a target variable, oil temperature.
21
+ * **`ECL`** The Electricity dataset includes hourly electricity consumption data from 370 clients, providing insights into consumer-level load patterns. Data is collected from 1st January, 2011 with a sampling interval of 15 minutes.
22
+ * **`Weather`** Weather dataset consists of one year of meteorological measurements recorded every 10 minutes across 21 weather stations of the Max Planck Biogeochemistry Institute in Germany. It includes 21 variables such as air temperature, humidity, and wind speed, etc.
23
+ * **`Exchange`** Exchange comprises daily exchange rate records from 1990 to 2016 for eight foreign currencies, including those of Australia, the United Kingdom, China, Japan, Canada, Singapore, Switzerland, and New Zealand. The data is sampled at a one-day interval.
24
+ * **`ILI`** The Influenza-like Illness (ILI) dataset captures the weekly number of reported cases involving severe influenza symptoms with complications.
25
+ * **`Electricity`** This dataset represents the hourly electricity consumption of 321 clients from 2012 to 2014, measured in kilowatts (kW). It was originally extracted from the UCI repository.
26
+ * **`Solar`** This dataset contains 137 time series representing hourly solar power production in the state of Alabama throughout 2006.
27
+ * **`Wind`** This dataset contains a single extensive daily time series representing wind power production (in megawatts) recorded at 4-second intervals starting from August 1, 2019. It was downloaded from the Australian Energy Market Operator (AEMO) online platform.
28
+ * **`Traffic`** This dataset contains 15 months of daily data (440 daily records) describing the occupancy rate (between 0 and 1) of different car lanes on San Francisco Bay Area freeways over time.
29
+ * **`Taxi`** This dataset contains spatio-temporal traffic time series of New York City taxi rides recorded at 1,214 locations every 30 minutes during January 2015 and January 2016.
30
+ * **`Pedestrian`** This dataset contains hourly pedestrian counts captured from 66 sensors in Melbourne starting from May 2009. The original dataset is regularly updated when new observations become available. The dataset used here contains pedestrian counts up to April 30, 2020.
31
+ * **`Air-Quality`** This dataset was used in the KDD Cup 2018 forecasting competition. It contains hourly air quality measurements from 59 stations in two cities: Beijing (35 stations) and London (24 stations) from January 1, 2017, to March 31, 2018. The air quality measurements include various metrics such as PM2.5, PM10, NO2, CO, O3, and SO2. Missing values were imputed using leading zeros or the Last Observation Carried Forward (LOCF) method.
32
+ * **`Temperature`** This dataset contains 32,072 daily time series with temperature observations and rain forecasts gathered by the Australian Bureau of Meteorology from 422 weather stations across Australia between May 2, 2015, and April 26, 2017. Missing values were replaced with zeros, and the mean temperature column was extracted for use.
33
+ * **`Rain`** This dataset focuses on rain data extracted from the same source as the Temperature dataset.
34
+ * **`NN5`** This dataset was used in the NN5 forecasting competition. It contains 111 time series from the banking domain with the goal of predicting daily cash withdrawals from ATMs in the UK. Missing values were replaced by the median across all corresponding days of the week throughout the entire series.
35
+ * **`Fred-MD`** This dataset contains 107 monthly time series reflecting various macroeconomic indicators sourced from the Federal Reserve Bank’s FRED-MD database. The series have been differenced and log-transformed following established practices in the literature.
36
+ * **`Web`** This dataset was used in the Kaggle Wikipedia Web Traffic forecasting competition. It contains 145,063 daily time series representing the number of hits or web traffic for Wikipedia pages from July 1, 2015, to September 10, 2017. Missing values were replaced with zeros.
37
+ * **`M4`** The M4 dataset comprises 100,000 time series from various domains, used in the fourth Makridakis forecasting competition (M4 Competition) organized by Spyros Makridakis. It includes series with multiple frequencies—yearly, quarterly, monthly, weekly, daily, and hourly—along with an Info file providing details such as series ID, category, frequency, forecast horizon, seasonal period, and starting date.
README.vi.md ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Tập Dữ Liệu Chuẩn Mực Dự Trắc Chuỗi Thời Gian
2
+
3
+ ## Ngôn Ngữ Tài Liệu
4
+
5
+ [**简体中文**](./README.zh-CN.md) | [**English**](./README.md) | [**Tiếng Việt**](./README.vi.md)
6
+
7
+ ## Tải Tập Dữ Liệu
8
+
9
+ * [**https://huggingface.co/datasets/Duyu/Time-Series-Forecasting-Benchmark-Datasets/tree/main**](https://huggingface.co/datasets/Duyu/Time-Series-Forecasting-Benchmark-Datasets/tree/main)
10
+
11
+ * [**https://github.com/duyu09/TimeSeries-Forecasting-Dataset/releases/download/v1.0.0/dataset.7z**](https://github.com/duyu09/TimeSeries-Forecasting-Dataset/releases/download/v1.0.0/dataset.7z)
12
+
13
+ ## Mô Tả Tập Dữ Liệu
14
+
15
+ * **`ETT`** Tập dữ liệu Electricity Transformer Temperature (ETT) là một chuẩn đánh giá quan trọng cho bài toán dự báo trong hệ thống điện. Tập dữ liệu bao gồm hai năm dữ liệu được thu thập từ hai huyện khác nhau tại Trung Quốc. Để phân tích ảnh hưởng của độ phân giải thời gian, Tập dữ liệu được chia thành bốn tập con với các tần suất lấy mẫu khác nhau: ETTh1 và ETTh2 được lấy mẫu theo chu kỳ 1 giờ, trong khi ETTm1 và ETTm2 được lấy mẫu theo chu kỳ 15 phút. Mỗi điểm dữ liệu bao gồm sáu đặc trưng liên quan đến tải điện và một biến mục tiêu là nhiệt độ dầu.
16
+ * **`ECL`** Tập dữ liệu Electricity bao gồm dữ liệu tiêu thụ điện theo giờ của 370 khách hàng, cung cấp cái nhìn chi tiết về các mô hình phụ tải ở cấp độ người dùng. Dữ liệu được thu thập từ ngày 1 tháng 1 năm 2011 với khoảng thời gian lấy mẫu là 15 phút.
17
+ * **`Weather`** Tập dữ liệu Weather bao gồm một năm các phép đo khí tượng được ghi nhận mỗi 10 phút tại 21 trạm thời tiết của Viện Sinh địa hóa Max Planck ở Đức. Tập dữ liệu bao gồm 21 biến, chẳng hạn như nhiệt độ không khí, độ ẩm và tốc độ gió, v.v.
18
+ * **`Exchange`** Tập dữ liệu Exchange bao gồm các bản ghi tỷ giá hối đoái hàng ngày từ năm 1990 đến năm 2016 cho tám loại tiền tệ nước ngoài, bao gồm Úc, Vương quốc Anh, Trung Quốc, Nhật Bản, Canada, Singapore, Thụy Sĩ và New Zealand. Dữ liệu được lấy mẫu theo chu kỳ một ngày.
19
+ * **`ILI`** Tập dữ liệu Influenza-like Illness (ILI) ghi nhận số ca bệnh được báo cáo hàng tuần liên quan đến các triệu chứng cúm nghiêm trọng kèm theo biến chứng.
20
+ * **`Electricity`** Tập dữ liệu này biểu diễn mức tiêu thụ điện theo giờ của 321 khách hàng trong giai đoạn từ năm 2012 đến năm 2014, được đo bằng kilowatt (kW). Dữ liệu ban đầu được trích xuất từ kho dữ liệu UCI.
21
+ * **`Solar`** Tập dữ liệu này bao gồm 137 chuỗi thời gian biểu diễn sản lượng điện mặt trời theo giờ tại bang Alabama trong suốt năm 2006.
22
+ * **`Wind`** Tập dữ liệu này bao gồm một chuỗi thời gian hàng ngày dài duy nhất, biểu diễn sản lượng điện gió (tính bằng megawatt) được ghi nhận với khoảng thời gian 4 giây, bắt đầu từ ngày 1 tháng 8 năm 2019. Dữ liệu được tải về từ nền tảng trực tuyến của Cơ Quan Thị Trường Năng Lượng Úc (AEMO).
23
+ * **`Traffic`** Tập dữ liệu này bao gồm 15 tháng dữ liệu hàng ngày (440 bản ghi theo ngày) mô tả tỷ lệ chiếm dụng (trong khoảng từ 0 đến 1) của các làn xe khác nhau trên hệ thống đường cao tốc khu vực Vịnh San Francisco theo thời gian.
24
+ * **`Taxi`** Tập dữ liệu này bao gồm các chuỗi thời gian giao thông không gian–thời gian của các chuyến taxi tại Thành phố New York, được ghi nhận tại 1.214 địa điểm mỗi 30 phút trong các tháng 1 năm 2015 và tháng 1 năm 2016.
25
+ * **`Pedestrian`** Tập dữ liệu này bao gồm số lượng người đi bộ được ghi nhận theo giờ từ 66 cảm biến tại Melbourne, bắt đầu từ tháng 5 năm 2009. Tập dữ liệu gốc được cập nhật định kỳ khi có quan sát mới. Tập dữ liệu được sử dụng ở đây bao gồm số liệu đến ngày 30 tháng 4 năm 2020.
26
+ * **`Air-Quality`** Tập dữ liệu này được sử dụng trong cuộc thi dự báo KDD Cup 2018. Nó bao gồm các phép đo chất lượng không khí theo giờ từ 59 trạm tại hai thành phố: Bắc Kinh (35 trạm) và London (24 trạm), trong khoảng thời gian từ ngày 1 tháng 1 năm 2017 đến ngày 31 tháng 3 năm 2018. Các phép đo chất lượng không khí bao gồm nhiều chỉ số như PM2.5, PM10, NO2, CO, O3 và SO2. Các giá trị bị thiếu được nội suy bằng cách điền số 0 ở đầu chuỗi hoặc sử dụng phương pháp Giá Trị Quan Sát Cuối Được Ghi Tiếp (LOCF).
27
+ * **`Temperature`** Tập dữ liệu này bao gồm 32.072 chuỗi thời gian hàng ngày với các quan sát nhiệt độ và dự báo mưa, được thu thập bởi Cục Khí tượng Úc từ 422 trạm thời tiết trên khắp nước Úc trong khoảng thời gian từ ngày 2 tháng 5 năm 2015 đến ngày 26 tháng 4 năm 2017. Các giá trị bị thiếu được thay thế bằng 0, và cột nhiệt độ trung bình được trích xuất để sử dụng.
28
+ * **`Rain`** Tập dữ liệu này tập trung vào dữ liệu lượng mưa được trích xuất từ cùng nguồn với Tập dữ liệu Temperature.
29
+ * **`NN5`** Tập dữ liệu này được sử dụng trong cuộc thi dự báo NN5. Nó bao gồm 111 chuỗi thời gian trong lĩnh vực ngân hàng, với mục tiêu dự báo lượng tiền mặt được rút hàng ngày từ các máy ATM tại Vương quốc Anh. Các giá trị bị thiếu được thay thế bằng giá trị trung vị của tất cả các ngày tương ứng trong tuần trên toàn bộ chuỗi.
30
+ * **`Fred-MD`** Tập dữ liệu này bao gồm 107 chuỗi thời gian hàng tháng phản ánh các chỉ số kinh tế vĩ mô khác nhau, được lấy từ cơ sở dữ liệu FRED-MD của Ngân hàng Dự trữ Liên bang Hoa Kỳ. Các chuỗi đã được lấy sai phân và biến đổi log theo các thông lệ đã được thiết lập trong tài liệu nghiên cứu.
31
+ * **`Web`** Tập dữ liệu này được sử dụng trong cuộc thi dự báo lưu lượng truy cập Wikipedia trên Kaggle. Nó bao gồm 145.063 chuỗi thời gian hàng ngày biểu diễn số lượt truy cập hoặc lưu lượng web của các trang Wikipedia, từ ngày 1 tháng 7 năm 2015 đến ngày 10 tháng 9 năm 2017. Các giá trị bị thiếu được thay thế bằng 0.
32
+ * **`M4`** Tập dữ liệu M4 bao gồm 100.000 chuỗi thời gian từ nhiều lĩnh vực khác nhau, được sử dụng trong cuộc thi dự báo Makridakis lần thứ tư (M4 Cuộc Thi) do Spyros Makridakis tổ chức. Tập dữ liệu bao gồm các chuỗi với nhiều tần suất khác nhau—hàng năm, hàng quý, hàng tháng, hàng tuần, hàng ngày và hàng giờ—cùng với một tệp Info cung cấp các thông tin như ID chuỗi, danh mục, tần suất, chân trời dự báo, chu kỳ mùa vụ và ngày bắt đầu.
33
+
README.zh-CN.md ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # 时间序列预测基准数据集
2
+
3
+ ## 文档语言
4
+
5
+ [**简体中文**](./README.zh-CN.md) | [**English**](./README.md) | [**Tiếng Việt**](./README.vi.md)
6
+
7
+ ## 数据集下载
8
+
9
+ * [**https://huggingface.co/datasets/Duyu/Time-Series-Forecasting-Benchmark-Datasets/tree/main**](https://huggingface.co/datasets/Duyu/Time-Series-Forecasting-Benchmark-Datasets/tree/main)
10
+
11
+ * [**https://github.com/duyu09/TimeSeries-Forecasting-Dataset/releases/download/v1.0.0/dataset.7z**](https://github.com/duyu09/TimeSeries-Forecasting-Dataset/releases/download/v1.0.0/dataset.7z)
12
+
13
+ ## 数据集描述
14
+
15
+ * **`ETT`** Electricity Transformer Temperature (ETT) 数据集是评估电力预测的关键基准。它包含来自中国两个不同县市的两年的数据。为分析时间粒度的影响,该数据集被分为四个具有不同采样频率的子集:ETTh1和ETTh2以1小时为间隔采样,而ETTm1和ETTm2以15分钟为间隔采样。每个数据点包含六个电力负荷相关特征以及一个目标变量,即油温。
16
+ * **`ECL`** Electricity数据集包含了370个客户的小时级用电量数据,为消费者层面的负荷模式提供了洞察。数据自2011年1月1日开始收集,采样间隔为15分钟。
17
+ * **`Weather`** Weather数据集包含德国马克斯·普朗克生物地球化学研究所21个气象站记录的为期一年的气象测量数据,每10分钟记录一次。它包含21个变量,如气温、湿度和风速等。
18
+ * **`Exchange`** Exchange数据集包含了1990年至2016年间八种外币的每日汇率记录,包括澳大利亚、英国、中国、日本、加拿大、新加坡、瑞士和新西兰的货币。数据采样间隔为一天。
19
+ * **`ILI`** Influenza-like Illness (ILI) 数据集记录了每周报告的伴有严重并发症的流感症状病例数。
20
+ * **`Electricity`** 该数据集代表了2012年至2014年间321个客户的小时级电力消耗(以千瓦为单位)。它最初提取自UCI知识库。
21
+ * **`Solar`** 该数据集包含137条时间序列,代表了2006年全年亚拉巴马州的小时级太阳能发电量。
22
+ * **`Wind`** 该数据集包含一条每日采样的长时间序列,代表自2019年8月1日开始以4秒间隔记录的风力发电功率(以兆瓦计)。该数据下载自澳大利亚能源市场运营商(AEMO)在线平台。
23
+ * **`Traffic`** 该数据集包含15个月的每日数据(440条每日记录),描述了旧金山湾区高速公路上不同车道的占用率(介于0和1之间)随时间的变化。
24
+ * **`Taxi`** 该数据集包含纽约市出租车出行的时空交通时间序列,于2015年1月和2016年1月期间在1,214个地点每30分钟记录一次。
25
+ * **`Pedestrian`** 该数据集包含自2009年5月开始从墨尔本66个传感器捕获的小时级行人计数。原始数据集在有新观测数据时定期更新。此处使用的数据集包含截至2020年4月30日的行人计数。
26
+ * **`Air-Quality`** 该数据集曾用于KDD Cup 2018预测竞赛。它包含2017年1月1日至2018年3月31日期间,来自北京(35个站点)和伦敦(24个站点)共59个站点的每小时空气质量测量数据。空气质量测量指标包括PM2.5、PM10、NO2、CO、O3和SO2等多种。缺失值已使用前导零或最后观测值结转法进行插补。
27
+ * **`Temperature`** 该数据集包含32,072条每日时间序列,记录了澳大利亚气象局在2015年5月2日至2017年4月26日期间从全国422个气象站收集的温度观测数据和降雨预测数据。缺失值已替换为零,并提取了平均温度列以供使用。
28
+ * **`Rain`** 该数据集聚焦于从与Temperature数据集相同来源提取的降雨数据。
29
+ * **`NN5`** 该数据集曾用于NN5预测竞赛。它包含来自银行业领域的111条时间序列,目标是预测英国ATM机的每日现金取款量。缺失值已用整个序列中对应星期几的中位数替换。
30
+ * **`Fred-MD`** 该数据集包含107条月度时间序列,反映了来自美联储FRED-MD数据库的各种宏观经济指标。这些序列已根据文献中的既定实践进行了差分和对数转换。
31
+ * **`Web`** 该数据集曾用于Kaggle维基百科网络流量预测竞赛。它包含145,063条每日时间序列,代表了2015年7月1日至2017年9月10日期间维基百科页面的点击量或网络流量。缺失值已替换为零。
32
+ * **`M4`** M4数据集包含来自多个领域的100,000条时间序列,用于由Spyros Makridakis组织的第四届Makridakis预测竞赛(M4竞赛)。它包括具有多种频率(年度、季度、月度、周度、每日和每小时)的序列,以及一个提供序列ID、类别、频率、预测范围、季节性周期和起始日期等详细信息的Info文件。