dvc.lock 3.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125
  1. schema: '2.0'
  2. stages:
  3. make_data:
  4. cmd: python3 make_dataset.py
  5. deps:
  6. - path: make_dataset.py
  7. md5: d700791fac216077c43596a5ea326f74
  8. outs:
  9. - path: remote://remote-wfs/rML-raw-data.csv
  10. md5: 15ae93ba7e913bcad890c2461048a9c9
  11. eval:
  12. cmd: python3 src/evaluate.py
  13. deps:
  14. - path: data/processed/rML-test.csv
  15. md5: 629a4d0505914f5599bb94be702341b5
  16. size: 2862132
  17. - path: models/model.pkl
  18. md5: 98f38f3ef55a5cfe9a0fb27a1f6616ae
  19. size: 1127
  20. - path: src/evaluate.py
  21. md5: 5f060c468f17fc53e9a57d2681cf291a
  22. size: 1590
  23. params:
  24. params.yaml:
  25. pre_process:
  26. chunk_size: 5000
  27. target_col: is_top_decile
  28. train:
  29. loss: log
  30. use_text_cols: false
  31. use_number_category_cols: true
  32. outs:
  33. - path: models/metrics/test.yaml
  34. md5: 85fa7115fb52ca48ffe9e8de88d3bee0
  35. size: 102
  36. preprocessing:
  37. cmd: python3 src/preprocess.py
  38. deps:
  39. - path: data/raw/rML-raw-data.csv
  40. md5: ce3a47e51025812d0e5d2e5914feadb7
  41. size: 7361822
  42. - path: general_params.yml
  43. md5: fd72d874a18159ce4a7855eb23ce4165
  44. size: 44
  45. - path: src/preprocess.py
  46. md5: 6b1cf3af3a8e62805f8d81f9e1dcd559
  47. size: 2620
  48. outs:
  49. - path: data/processed/rML-test.csv
  50. md5: 629a4d0505914f5599bb94be702341b5
  51. size: 2862132
  52. - path: data/processed/rML-train.csv
  53. md5: 9b1f377458cc33aaa5b72a63fbb757fe
  54. size: 8087263
  55. training:
  56. cmd: python3 src/training.py
  57. deps:
  58. - path: data/processed/rML-train.csv
  59. md5: 9b1f377458cc33aaa5b72a63fbb757fe
  60. size: 8087263
  61. - path: general_params.yml
  62. md5: fd72d874a18159ce4a7855eb23ce4165
  63. size: 44
  64. - path: src/model_def.py
  65. md5: f93783b7270ebd57d40f9f6c1b8e4188
  66. size: 2174
  67. - path: src/training.py
  68. md5: 388970c00f0c5eb6b27a9077a55584d0
  69. size: 871
  70. outs:
  71. - path: models/metrics/train.yaml
  72. md5: c4fb83687d866968d411df61c68b5d63
  73. size: 103
  74. - path: models/model.pkl
  75. md5: 98f38f3ef55a5cfe9a0fb27a1f6616ae
  76. size: 1127
  77. - path: params.yml
  78. md5: 2425d4cc306ef245b718ef1179191bf5
  79. size: 542
  80. pre_process:
  81. cmd: python3 src/pre_process.py
  82. deps:
  83. - path: data/raw/rML-raw-data.csv
  84. md5: ce3a47e51025812d0e5d2e5914feadb7
  85. size: 7361822
  86. - path: src/pre_process.py
  87. md5: e05a7f042ab6e3f81eb27e723a9668cf
  88. size: 2622
  89. params:
  90. params.yaml:
  91. pre_process:
  92. chunk_size: 5000
  93. target_col: is_top_decile
  94. outs:
  95. - path: data/processed/rML-test.csv
  96. md5: 629a4d0505914f5599bb94be702341b5
  97. size: 2862132
  98. - path: data/processed/rML-train.csv
  99. md5: 9b1f377458cc33aaa5b72a63fbb757fe
  100. size: 8087263
  101. train:
  102. cmd: python3 src/train.py
  103. deps:
  104. - path: data/processed/rML-train.csv
  105. md5: 9b1f377458cc33aaa5b72a63fbb757fe
  106. size: 8087263
  107. - path: src/model_def.py
  108. md5: 2f20bd7e1336a86ad3adff8b859ccffb
  109. size: 1985
  110. - path: src/train.py
  111. md5: 8bd59fb472442f2b7851ee1e25da45cb
  112. size: 750
  113. params:
  114. params.yaml:
  115. train:
  116. loss: log
  117. use_text_cols: false
  118. use_number_category_cols: true
  119. outs:
  120. - path: models/metrics/train.yaml
  121. md5: c4fb83687d866968d411df61c68b5d63
  122. size: 103
  123. - path: models/model.pkl
  124. md5: 98f38f3ef55a5cfe9a0fb27a1f6616ae
  125. size: 1127
Tip!

Press p or to see the previous file or, n or to see the next file