Hugging Face's logo

Datasets:
THULab
/
rossmann

Tasks:
Time Series Forecasting
Modalities:
Time-series
Sub-tasks:
univariate-time-series-forecasting
multivariate-time-series-forecasting
Size:
1K<n<10K
Tags:
forecasting
benchmark
fev
arxiv:2509.26468
tsfile
modality:timeseries
License:
Dataset card Files
xet
 Community
VGalaxies666 commited on
Commit
d6aad5c
·
verified ·
1 Parent(s): 2842d0f

chore: sync metadata, drop original files

Browse files
Files changed (1) hide show
  1. README.md +118 -114
README.md CHANGED
@@ -1,114 +1,118 @@
1
- ---
2
- license: other
3
- task_categories:
4
- - time-series-forecasting
5
- task_ids:
6
- - univariate-time-series-forecasting
7
- - multivariate-time-series-forecasting
8
- annotations_creators:
9
- - no-annotation
10
- source_datasets:
11
- - original
12
- tags:
13
- - forecasting
14
- - benchmark
15
- - fev
16
- - arxiv:2509.26468
17
- - tsfile
18
- - modality:timeseries
19
- size_categories:
20
- - 1K<n<10K
21
- pretty_name: rossmann (TsFile format)
22
- configs:
23
- - config_name: default
24
- data_files:
25
- - split: train
26
- path: "**/*.tsfile"
27
- modality:
28
- - timeseries
29
- ---
30
-
31
- # rossmann (TsFile format)
32
-
33
- This repository contains time-series forecasting data stored in [Apache TsFile](https://tsfile.apache.org/) format.
34
-
35
- ## Summary
36
-
37
- - FEV subset: `rossmann`
38
- - Unified source collection: [`autogluon/fev_datasets`](https://huggingface.co/datasets/autogluon/fev_datasets)
39
- - Original source: https://www.kaggle.com/competitions/rossmann-store-sales
40
- - Paper / citation: [[21]](https://www.kaggle.com/competitions/rossmann-store-sales/overview/citation)
41
- - Series: 1,115
42
- - Modalities: Time-series
43
- - TsFile rows (flattened observations): 8,242,080
44
- - Frequencies: 1D, 1W
45
- - TsFile files: 3
46
- - Time precision: milliseconds (`INT64`).
47
-
48
- Licensing and citation requirements follow the original source. This repository does not claim ownership of the original data.
49
-
50
- ## Dataset Statistics
51
-
52
- | Frequency | Series | Median series length | TsFile rows (observations) | Dynamic columns | Static columns | Data files |
53
- |---|---:|---:|---:|---:|---:|---|
54
- | 1D | 1,115 | 942 | 7,352,310 | 7 | 10 | `1D/1D_1..1D_2.tsfile` (2 shards) |
55
- | 1W | 1,115 | 133 | 889,770 | 6 | 10 | `1W/1W.tsfile` |
56
-
57
- ## Files
58
-
59
- The Hugging Face dataset card YAML points `configs.data_files` to all `*.tsfile` files in this repository.
60
-
61
- - `1D/1D_1.tsfile`
62
- - `1D/1D_2.tsfile`
63
- - `1W/1W.tsfile`
64
-
65
- ## TsFile Storage Model
66
-
67
- - Each original series (`id`) is stored as one TsFile device.
68
- - Static covariate columns are stored as TAG columns: `Store, StoreType, Assortment, CompetitionDistance, CompetitionOpenSinceMonth, CompetitionOpenSinceYear, Promo2, Promo2SinceWeek, Promo2SinceYear, PromoInterval`.
69
- - Time-varying targets and dynamic covariates are stored as FIELD measurements.
70
- - Source `timestamp` values are mapped to the TsFile `Time` column as millisecond timestamps.
71
- - Table name(s): rossmann_1D, rossmann_1W.
72
-
73
- ### Column Schema
74
-
75
- | Column | Role | TsFile type |
76
- |---|---|---|
77
- | `Time` | Time column | INT64 |
78
- | `id` | TAG (device dimension) | STRING |
79
- | `Store` | TAG (device dimension) | DOUBLE |
80
- | `StoreType` | TAG (device dimension) | STRING |
81
- | `Assortment` | TAG (device dimension) | STRING |
82
- | `CompetitionDistance` | TAG (device dimension) | DOUBLE |
83
- | `CompetitionOpenSinceMonth` | TAG (device dimension) | DOUBLE |
84
- | `CompetitionOpenSinceYear` | TAG (device dimension) | DOUBLE |
85
- | `Promo2` | TAG (device dimension) | DOUBLE |
86
- | `Promo2SinceWeek` | TAG (device dimension) | DOUBLE |
87
- | `Promo2SinceYear` | TAG (device dimension) | DOUBLE |
88
- | `PromoInterval` | TAG (device dimension) | STRING |
89
- | `DayOfWeek` | FIELD (measurement) | FLOAT |
90
- | `Sales` | FIELD (measurement) | FLOAT |
91
- | `Customers` | FIELD (measurement) | FLOAT |
92
- | `Open` | FIELD (measurement) | FLOAT |
93
- | `Promo` | FIELD (measurement) | FLOAT |
94
- | `StateHoliday` | FIELD (measurement) | STRING |
95
- | `SchoolHoliday` | FIELD (measurement) | FLOAT |
96
-
97
- > Note: 2230 original `id` values contained invalid identifier characters and were normalized to valid device names, for example 1→_1, 2→_2, 3→_3.
98
-
99
- ## Conversion Notes
100
-
101
- - The source FEV format stores each time series as one nested row containing `id`, `timestamp[]`, and target or covariate arrays.
102
- - The TsFile conversion flattens those nested arrays into long rows. Therefore, the `TsFile rows` values above correspond to the number of timestamped observations after flattening.
103
- - TAG columns identify the device and static metadata. FIELD columns contain values that change over time.
104
- - Large logical tables may be split into multiple `.tsfile` shards such as `<name>_1.tsfile`, `<name>_2.tsfile`, and so on. Shards listed for the same frequency belong to the same logical table.
105
-
106
- ## Reading Example
107
-
108
- ```python
109
- from tsfile import TsFileReader
110
-
111
- reader = TsFileReader("1D/1D_1.tsfile")
112
- schemas = reader.get_all_table_schemas()
113
- # Table name(s): rossmann_1D, rossmann_1W
114
- ```
 
 
 
 
 
1
+ ---
2
+ license: other
3
+ task_categories:
4
+ - time-series-forecasting
5
+ task_ids:
6
+ - univariate-time-series-forecasting
7
+ - multivariate-time-series-forecasting
8
+ annotations_creators:
9
+ - no-annotation
10
+ source_datasets:
11
+ - original
12
+ tags:
13
+ - forecasting
14
+ - benchmark
15
+ - fev
16
+ - arxiv:2509.26468
17
+ - tsfile
18
+ - modality:timeseries
19
+ - timeseries
20
+ - format:tsfile
21
+ size_categories:
22
+ - 100K<n<1M
23
+ pretty_name: rossmann (TsFile format)
24
+ configs:
25
+ - config_name: default
26
+ data_files:
27
+ - split: train
28
+ path: "**/*.tsfile"
29
+ modality:
30
+ - tabular
31
+ - text
32
+ - timeseries
33
+ ---
34
+
35
+ # rossmann (TsFile format)
36
+
37
+ This repository contains time-series forecasting data stored in [Apache TsFile](https://tsfile.apache.org/) format.
38
+
39
+ ## Summary
40
+
41
+ - FEV subset: `rossmann`
42
+ - Unified source collection: [`autogluon/fev_datasets`](https://huggingface.co/datasets/autogluon/fev_datasets)
43
+ - Original source: https://www.kaggle.com/competitions/rossmann-store-sales
44
+ - Paper / citation: [[21]](https://www.kaggle.com/competitions/rossmann-store-sales/overview/citation)
45
+ - Series: 1,115
46
+ - Modalities: Time-series
47
+ - TsFile rows (flattened observations): 8,242,080
48
+ - Frequencies: 1D, 1W
49
+ - TsFile files: 3
50
+ - Time precision: milliseconds (`INT64`).
51
+
52
+ Licensing and citation requirements follow the original source. This repository does not claim ownership of the original data.
53
+
54
+ ## Dataset Statistics
55
+
56
+ | Frequency | Series | Median series length | TsFile rows (observations) | Dynamic columns | Static columns | Data files |
57
+ |---|---:|---:|---:|---:|---:|---|
58
+ | 1D | 1,115 | 942 | 7,352,310 | 7 | 10 | `1D/1D_1..1D_2.tsfile` (2 shards) |
59
+ | 1W | 1,115 | 133 | 889,770 | 6 | 10 | `1W/1W.tsfile` |
60
+
61
+ ## Files
62
+
63
+ The Hugging Face dataset card YAML points `configs.data_files` to all `*.tsfile` files in this repository.
64
+
65
+ - `1D/1D_1.tsfile`
66
+ - `1D/1D_2.tsfile`
67
+ - `1W/1W.tsfile`
68
+
69
+ ## TsFile Storage Model
70
+
71
+ - Each original series (`id`) is stored as one TsFile device.
72
+ - Static covariate columns are stored as TAG columns: `Store, StoreType, Assortment, CompetitionDistance, CompetitionOpenSinceMonth, CompetitionOpenSinceYear, Promo2, Promo2SinceWeek, Promo2SinceYear, PromoInterval`.
73
+ - Time-varying targets and dynamic covariates are stored as FIELD measurements.
74
+ - Source `timestamp` values are mapped to the TsFile `Time` column as millisecond timestamps.
75
+ - Table name(s): rossmann_1D, rossmann_1W.
76
+
77
+ ### Column Schema
78
+
79
+ | Column | Role | TsFile type |
80
+ |---|---|---|
81
+ | `Time` | Time column | INT64 |
82
+ | `id` | TAG (device dimension) | STRING |
83
+ | `Store` | TAG (device dimension) | DOUBLE |
84
+ | `StoreType` | TAG (device dimension) | STRING |
85
+ | `Assortment` | TAG (device dimension) | STRING |
86
+ | `CompetitionDistance` | TAG (device dimension) | DOUBLE |
87
+ | `CompetitionOpenSinceMonth` | TAG (device dimension) | DOUBLE |
88
+ | `CompetitionOpenSinceYear` | TAG (device dimension) | DOUBLE |
89
+ | `Promo2` | TAG (device dimension) | DOUBLE |
90
+ | `Promo2SinceWeek` | TAG (device dimension) | DOUBLE |
91
+ | `Promo2SinceYear` | TAG (device dimension) | DOUBLE |
92
+ | `PromoInterval` | TAG (device dimension) | STRING |
93
+ | `DayOfWeek` | FIELD (measurement) | FLOAT |
94
+ | `Sales` | FIELD (measurement) | FLOAT |
95
+ | `Customers` | FIELD (measurement) | FLOAT |
96
+ | `Open` | FIELD (measurement) | FLOAT |
97
+ | `Promo` | FIELD (measurement) | FLOAT |
98
+ | `StateHoliday` | FIELD (measurement) | STRING |
99
+ | `SchoolHoliday` | FIELD (measurement) | FLOAT |
100
+
101
+ > Note: 2230 original `id` values contained invalid identifier characters and were normalized to valid device names, for example 1→_1, 2→_2, 3→_3.
102
+
103
+ ## Conversion Notes
104
+
105
+ - The source FEV format stores each time series as one nested row containing `id`, `timestamp[]`, and target or covariate arrays.
106
+ - The TsFile conversion flattens those nested arrays into long rows. Therefore, the `TsFile rows` values above correspond to the number of timestamped observations after flattening.
107
+ - TAG columns identify the device and static metadata. FIELD columns contain values that change over time.
108
+ - Large logical tables may be split into multiple `.tsfile` shards such as `<name>_1.tsfile`, `<name>_2.tsfile`, and so on. Shards listed for the same frequency belong to the same logical table.
109
+
110
+ ## Reading Example
111
+
112
+ ```python
113
+ from tsfile import TsFileReader
114
+
115
+ reader = TsFileReader("1D/1D_1.tsfile")
116
+ schemas = reader.get_all_table_schemas()
117
+ # Table name(s): rossmann_1D, rossmann_1W
118
+ ```