Create README.md

README.md

@@ -0,0 +1,45 @@
+# muvr-training-data
+
+This repository is a collection of training datasets used for analysis and model training during model evaluation. 
+Newly added datasets should be listed here including there format and a short description.
+
+Some of the datasets might be zipped since they would be to big otherwise.
+
+## (partial) List of available datasets
+
+#### combined
+
+| name                                         	| format                 	| zipped 	| split    	| comment                                                                	|
+|----------------------------------------------	|------------------------	|:------:	|----------	|------------------------------------------------------------------------	|
+| 09-09-15-activity-slacking.zip      	| combined-1200          	|    x   	| presplit 	| Collection of movement data (walking, still, other) vs. exercise data. 	|
+| 18-09-15-triceps-biceps-lateral.zip 	| single-exercise-simple 	|    x   	|          	| Collection of three exercises.                                         	|
+| combined-tbl-activity-slacking.zip  	| single-exercise-simple 	|    x   	| presplit 	| Combination of 09-09-15-activity-slacking.zip and                      	|
+
+#### labelled
+Jan's collected exercises. Contains `raw-acceleration` and `single-exercise-extended` data files. Not all `raw` files have been converted.
+
+#### raw
+Unlabeled data in `raw-acceleration` format.
+
+## Format descriptions
+
+#### raw-acceleration
+The dataset is a collection of **unlabeled** acceleration data, one file per exercise. A file for an exercise contains dumped sensor data in an array like format
+
+#### single-exercise-simple
+The dataset is a collection of **labeled** data, one file per exercise. A file for an exercise is in CSV format and there is one row per measurement:
+```
+exercise group | exercise name | x | y | z |
+```
+
+#### single-exercise-extended
+The dataset is a collection of **labeled** data, one file per exercise. A file for an exercise is in CSV format and there is one row per measurement:
+```
+exercise group | exercise name | <unknown> | intensity? | weight | x | y | z |
+```
+
+#### combined-1200
+The dataset is a collection of **labeled** data in a single file. The file is in CSV and there is one row per example:
+```
+label | X1 | X2 | ... | X400  | Y1 | Y2 | ... | Y400   | Z1 | Z2 | ... | Z400 
+```