- Numpy (Ver: 1.14.3)
- Matplotlib (Ver: 2.2.2)
- Astropy (Ver: 2.0.7)
- Sklearn (Ver: 0.19.1)
- TQDM (Ver: 4.23.0)
- PyLMNN (Ver: 1.5.2)
- Metric_Learn (Ver: 0.3.0)
- Using arguments:
- -a | --algorithm. The type of ML to use. Options: 0 for kNN, 1 for Random Forest
- -t | --testType. The training/testing regime to use. Options: 0 for normal, 1 for sub-field test with train = ELAIS-S1, 2 = sub-field test with train = eCDFS
- -d | --distType. The distance metric to use with the kNN Algorithm. Options: 1 for Manhattan, 2 for Euclidean, 99 for Mahalanobis.
- Note. Use the Euclidean Distance Metric when using a learned distance metric.
- -b | --bootstrapSize. Number of bootstrap intervals. Do not use if you don't want bootstrapped error bars.
- -c | --classification. Whether to use Classification-based algorithms. Any value will trigger the classification mode of the algorithm chosen.
- -m | --metricLearn. Should metric learning be used? Don't enter for no.
- -k | --kNeighbours. Number of neighbours to use.
- Note. Using this option means the value of k used by the kNN algorithm or the number of trees used by the Random Forest will not be chosen based on the lowest outlier rate.
Redshift.py -a 0 -t 0 -d 2- Using this will run the kNN algorithm in regression mode, using a random training sample and the Euclidean Distance Metric
Redshift.py -a 0 -t 1 -d 2 -c True- Using this will run the kNN algorithm in classification mode, using the ELAIS-S1 field as a training set and the Euclidean Distance Metric
Redshift.py -a 0 -t 2 -d 99- Using this will run the kNN algorithm in regression mode, using the eCDFS field as a training set, and the Mahalanobis distance metric.
Redshift.py -a 0 -t 0 -d 2 -m True -b 1000- Using this will run the kNN algorithm in regression mode, using a random training sample and the Euclidean Distance metric after transforming the data using the MLKR distance metric. 95% confidence intervals will be computed using 1000 bootstrap samples.
Redshift.py -a 0 -t 0 -d 2 -m True -c True- Using this will run the kNN algorithm in classification mode, using a random training sample and the Euclidean Distance metric after transforming the data using the LMNN distance metric.
Redshift.py -a 1 -t 0- Using this will run the Random Forest algorithm in regression mode, using a random training sample.
Redshift.py -a 1 -t 0 -c True- Using this will run the Random Forest algorithm in classification mode, using a random training sample.