Datasets:

ArXiv:
License:
gorold commited on
Commit
b822614
1 Parent(s): 4087379

update README

Browse files
Files changed (1) hide show
  1. README.md +32 -18
README.md CHANGED
@@ -14,15 +14,14 @@ size_categories:
14
  Datasets accompanying the paper "Pushing the Limits of Pre-training for Time Series Forecasting in the CloudOps Domain".
15
 
16
  ## Quick Start
 
 
17
  ```python
18
  from datasets import load_dataset
19
 
20
  dataset = load_dataset('Salesforce/cloudops_tsf', 'azure_vm_traces_2017')
21
- ```
22
 
23
- ## Available Datasets
24
- ### azure_vm_traces_2017
25
- ```python
26
  DatasetDict({
27
  train_test: Dataset({
28
  features: ['start', 'target', 'item_id', 'feat_static_cat', 'feat_static_real', 'past_feat_dynamic_real'],
@@ -37,6 +36,9 @@ DatasetDict({
37
 
38
  ### borg_cluster_data_2011
39
  ```python
 
 
 
40
  DatasetDict({
41
  train_test: Dataset({
42
  features: ['start', 'target', 'item_id', 'feat_static_cat', 'past_feat_dynamic_real'],
@@ -51,6 +53,9 @@ DatasetDict({
51
 
52
  ### alibaba_cluster_trace_2018
53
  ```python
 
 
 
54
  DatasetDict({
55
  train_test: Dataset({
56
  features: ['start', 'target', 'item_id', 'feat_static_cat', 'past_feat_dynamic_real'],
@@ -71,7 +76,7 @@ config = load_dataset_builder('Salesforce/cloudops_tsf', 'azure_vm_traces_2017')
71
  print(config)
72
 
73
  CloudOpsTSFConfig(
74
- name='alibaba_cluster_trace_2018',
75
  version=1.0.0,
76
  data_dir=None,
77
  data_files=None,
@@ -79,23 +84,32 @@ CloudOpsTSFConfig(
79
  prediction_length=48,
80
  freq='5T',
81
  stride=48,
82
- univariate=False,
83
- multivariate=True,
84
- optional_fields=('feat_static_cat', 'past_feat_dynamic_real'),
 
 
 
 
85
  rolling_evaluations=12,
86
- test_split_date=Period('2018-01-08 11:55', '5T'),
87
  _feat_static_cat_cardinalities={
88
  'pretrain': (
89
- ('container_id', 64457),
90
- ('app_du', 9484)),
 
 
 
91
  'train_test': (
92
- ('container_id', 6048),
93
- ('app_du', 1292)
 
 
94
  )
95
  },
96
- target_dim=2,
97
- feat_static_real_dim=0,
98
- past_feat_dynamic_real_dim=6
99
  )
100
  ```
101
  ```test_split_date``` is provided to achieve the same train-test split as given in the paper.
@@ -117,11 +131,11 @@ The datasets were processed from the following original sources. Please cite the
117
  * https://github.com/alibaba/clusterdata
118
 
119
  ## Citation
120
- ```
121
  @article{woo2023pushing,
122
  title={Pushing the Limits of Pre-training for Time Series Forecasting in the CloudOps Domain},
123
  author={Woo, Gerald and Liu, Chenghao and Kumar, Akshat and Sahoo, Doyen},
124
  journal={arXiv preprint arXiv:2310.05063},
125
  year={2023}
126
  }
127
- ```
 
14
  Datasets accompanying the paper "Pushing the Limits of Pre-training for Time Series Forecasting in the CloudOps Domain".
15
 
16
  ## Quick Start
17
+
18
+ ### azure_vm_traces_2017
19
  ```python
20
  from datasets import load_dataset
21
 
22
  dataset = load_dataset('Salesforce/cloudops_tsf', 'azure_vm_traces_2017')
23
+ print(dataset)
24
 
 
 
 
25
  DatasetDict({
26
  train_test: Dataset({
27
  features: ['start', 'target', 'item_id', 'feat_static_cat', 'feat_static_real', 'past_feat_dynamic_real'],
 
36
 
37
  ### borg_cluster_data_2011
38
  ```python
39
+ dataset = load_dataset('Salesforce/cloudops_tsf', 'borg_cluster_data_2011')
40
+ print(dataset)
41
+
42
  DatasetDict({
43
  train_test: Dataset({
44
  features: ['start', 'target', 'item_id', 'feat_static_cat', 'past_feat_dynamic_real'],
 
53
 
54
  ### alibaba_cluster_trace_2018
55
  ```python
56
+ dataset = load_dataset('Salesforce/cloudops_tsf', 'alibaba_cluster_trace_2018')
57
+ print(dataset)
58
+
59
  DatasetDict({
60
  train_test: Dataset({
61
  features: ['start', 'target', 'item_id', 'feat_static_cat', 'past_feat_dynamic_real'],
 
76
  print(config)
77
 
78
  CloudOpsTSFConfig(
79
+ name='azure_vm_traces_2017',
80
  version=1.0.0,
81
  data_dir=None,
82
  data_files=None,
 
84
  prediction_length=48,
85
  freq='5T',
86
  stride=48,
87
+ univariate=True,
88
+ multivariate=False,
89
+ optional_fields=(
90
+ 'feat_static_cat',
91
+ 'feat_static_real',
92
+ 'past_feat_dynamic_real'
93
+ ),
94
  rolling_evaluations=12,
95
+ test_split_date=Period('2016-12-13 15:55', '5T'),
96
  _feat_static_cat_cardinalities={
97
  'pretrain': (
98
+ ('vm_id', 177040),
99
+ ('subscription_id', 5514),
100
+ ('deployment_id', 15208),
101
+ ('vm_category', 3)
102
+ ),
103
  'train_test': (
104
+ ('vm_id', 17568),
105
+ ('subscription_id', 2713),
106
+ ('deployment_id', 3255),
107
+ ('vm_category', 3)
108
  )
109
  },
110
+ target_dim=1,
111
+ feat_static_real_dim=3,
112
+ past_feat_dynamic_real_dim=2
113
  )
114
  ```
115
  ```test_split_date``` is provided to achieve the same train-test split as given in the paper.
 
131
  * https://github.com/alibaba/clusterdata
132
 
133
  ## Citation
134
+ <pre>
135
  @article{woo2023pushing,
136
  title={Pushing the Limits of Pre-training for Time Series Forecasting in the CloudOps Domain},
137
  author={Woo, Gerald and Liu, Chenghao and Kumar, Akshat and Sahoo, Doyen},
138
  journal={arXiv preprint arXiv:2310.05063},
139
  year={2023}
140
  }
141
+ </pre>