1.Preparation for learning
1.1.Definition of learning themeName | Type | URL | JsonParam | QueryParam |
---|---|---|---|---|
01105004 |
post | |||
Add a new Machine Learning Theme. (language=0:English, 1:Japanese, 2:Other) (trainType=1:NaivebayesTextClassification, 2:LinearRegression, 3:LogisticRegression, 4:TimeSeriesAR, 5:TimeSeriesMA, 6:TimeSeriesARMA, 7:TimeSeriesARIMA) It is also possible to add the learning theme together with its item information. - For NaivebayesTextClassification: category is used here, and multiple items can be saved in the same category. - For Linear Regression Prediction: at least two items are required. Put a numerical string in trainText that is separated by spaces. The splitting count of every string should be identical. The numerical string in the last trainText is the true value. - For Logistic Regression Classification: at least two items are required. The classification values are stored in the last trainText, and they can be character strings. These numerical strings or character strings in trainText are separated by spaces, and the splitting count of every string should be identical. - For Time Series Prediction: only a trainText is required. Put a numerical string in trainText that is separated by spaces. If successful, please record the trainKey and resultKey. |
||||
01105005 |
put | |||
Update the property information for a Learning Theme. (language=0: English, 1: Japanese, 2: Other) (trainType=1: Naive Bayes Text Classification, 2: Linear Regression, 3: Logistic Regression, 4: Time Series AR, 5: Time Series MA, 6: Time Series ARMA, 7: Time Series ARIMA) It is also possible to update the Learning Theme's item information. - For Naive Bayes Text Classification: category is required to be input here, and multiple items can be saved in the same category. - For Linear Regression Prediction: at least two items are required. Put a numerical string in trainText that is separated with spaces. The splitting count of every string should be identical. The numerical string in the last trainText is the true value. - For Logistic Regression Classification: at least two items are required. The classification values are stored in the last trainText, and they can be character strings or numerical strings. These strings in trainText should be separated with spaces, and the splitting count of every string should be identical. - For Time Series Prediction: only a trainText is required. Put a numerical string in trainText that is separated with spaces. |
||||
01105003 |
get | |||
Get the property information of Learning Theme including resultKey. |
||||
01105006 |
delete | |||
delete the Learning Theme. |
||||
01105007 |
put | |||
Set the status of Learning Theme.(StatusId=0:new; 1:validate; 2:logical deletion) Important: please set [StatusId] to [1] in advance before calling the following Learning/Analyse/Prediction API. |
||||
1.2.Input of learning data
Name | Type | URL | JsonParam | QueryParam |
---|---|---|---|---|
01106004 |
post | |||
Add new items to a specified Learning Theme. - For NaivebayesTextClassification: category is required to be input here, and multiple items can be saved in the same category. - For Linear Regression Prediction: at least two items are required. Put a numerical string in trainText which is separated with spaces. The splitting count of every string must be identical. The numerical string in the last trainText is the true value. - For Logistic Regression Classification: at least two items are required. The classification values are stored in the last trainText, and they can be character strings or numerical strings. These strings in trainText are separated with spaces, and the splitting count of every string must be identical. - For Time Series Prediction: only a trainText is required. Put a numerical string in trainText which is separated with spaces. |
||||
01106005 |
put | |||
Update certain items of a specified Learning Theme. - For NaivebayesTextClassification: category is required to be input here, and multiple items can be saved in the same category. - For Linear Regression Prediction: at least two items are required. Put a numerical string in trainText which is separated by spaces, and the splitting count of every string is identical. The numerical string in the last trainText is the true value. - For Logistic Regression Classification: at least two items are required. The classification values are stored in the last trainText, and they can be character strings or numerical strings. These strings in trainText are separated by spaces, and the splitting count of every string is identical. - For Time Series Prediction: only a trainText is required. Put a numerical string in trainText which is separated by spaces. |
||||
01106001 |
get | |||
Get the items list of a specified Learning Theme. |
||||
01106002 |
get | |||
Get the CSV file of items for a specified Learning Theme. |
||||
01106003 |
get | |||
Get an Item property of a specified Learning Theme. |
||||
01106006 |
delete | |||
Delete Items of a specified Learning Theme. Multiple itemNos can be inputted. |
||||
2.Learning process(Generate Classification/Prediction Model)
Name | Type | URL | JsonParam | QueryParam |
---|---|---|---|---|
01105008 |
get | |||
Generate a Naive Bayes Classifier Model based on the text datasets provided as input in the Learning item. The process may take some time depending on the amount of learning items. |
||||
01105010 |
get | |||
Generate a Linear Single or Multiple Regression Prediction Model according to the trainText(Space-separated numeric string datasets) which is inputted in the Learning item in advance. The time taken to generate the model may vary depending on the amount of learning items. In the returned result list, if the coefficient of determination (Score) in the last row is close to 1, a satisfactory prediction will be available. (learnCount: number of learning times; learnRate: learning increment rate; interval: the output interval of the learning result) |
||||
01105012 |
post | |||
Generate a Logistic Regression Classification Model according to the trainText(Space-separated numeric string datasets) which was inputted in the Learning item in advance. The test data may use the input JSON here or a part of the trainText(numerical datasets) of the Learning item with testPercent(test data ratio). In the returned result list, if the Score in the last row is close to 1, a satisfying calculation accuracy will be available. (testPercent: test data ratio in the input training data set; learnCount: learning times; learnRate: learning increment rate; interval: the output interval of the learning result) |
||||
01105014 |
get | |||
Generate an Autoregressive Prediction Model according to the trainText (a space-separated numeric string dataset) which was inputted in the Learning item in advance. In the returned result list, if the Coefficient of Determination (Score) in the last row is close to 1, a satisfying prediction will be available. (learnCount: number of learning iterations; learnRate: learning increment rate; interval: output interval of learning result; pOrder: autoregressive times for learning) |
||||
01105016 |
get | |||
Generate a Moving Average Regression Prediction Model according to the trainText(a space-separated numeric string dataset) which is inputted in the Learning item in advance. In the returned result list, if the Coefficient of Determination (Score) in the last row is close to 1, a satisfying prediction will be available. (learnCount: learning times; learnRate: learning increment rate; interval: the output interval of the learning result; qOrder: Moving Average Regression times for Learning) |
||||
01105018 |
get | Generate an Autoregressive Moving Average Prediction Model based on the "trainText" dataset, which is a space-separated numeric string that has been previously inputted in the "Learning" section. In the returned result list, if the Coefficient of Determination (Score) in the last row is close to 1, it indicates that a satisfactory prediction is available. (learnCount: number of times to learn; learnRate: learning rate increment; interval: output interval of learning results; pOrder: number of times for Autoregressive learning; qOrder: number of times for Moving Average Regression learning) |
||
01105020 |
get | |||
Generate an Autoregressive Integrated Moving Average Prediction Model based on the "trainText" dataset, which is a space-separated numeric string that has been previously inputted in the "Learning" section. In the returned result list, if the Coefficient of Determination (Score) in the last row is close to 1, it indicates that a satisfactory prediction is available. (learnCount: number of times to learn; learnRate: learning rate increment; interval: output interval of learning results; pOrder: number of times for Autoregressive learning; dOrder: number of difference floors for learning; qOrder: number of times for Moving Average Regression learning) |
||||
3.Obtaining classification / prediction results
Name | Type | URL | JsonParam | QueryParam |
---|---|---|---|---|
01105009 |
post | |||
Enter text to get the analysis results of Naive Bayes Classifier Model. The analysis may take some time depending on the number of learning items. |
||||
01105011 |
post | |||
If you enter a numerical array, the prediction result of Linear Regression will be available. |
||||
01105013 |
post | |||
If you enter a numerical array, the probability result of Logistic Regression Classification will be available. |
||||
01105015 |
get | |||
Performing an Autoregressive prediction process will make the prediction result available. (outStartNo: starting number of the returned prediction result; outEndNo: ending number of the returned prediction result; cycleStartNo: starting number of the cycle in the learning data; cyclePeriod: cycle period in the learning data) |
||||
01105017 |
get | |||
Performing a Moving Average prediction process will make the prediction result available. (outStartNo: starting number of the returned prediction result; outEndNo: ending number of the returned prediction result; cycleStartNo: starting number of the cycle in the learning data; cyclePeriod: cycle period in the learning data) |
||||
01105019 |
get | |||
Performing an Autoregressive moving average prediction process will make the prediction result available. (outStartNo: starting number of the returned prediction result; outEndNo: ending number of the returned prediction result; cycleStartNo: starting number of the cycle in the learning data; cyclePeriod: cycle period in the learning data) |
||||
01105021 |
get | |||
Performing an Autoregressive integrated moving average prediction process will make the prediction result available. (outStartNo: starting number of the returned prediction result; outEndNo: ending number of the returned prediction result; cycleStartNo: starting number of the cycle in the learning data; cyclePeriod: cycle period in the learning data) |
||||
Statistical tool
Name | Type | URL | JsonParam | QueryParam |
---|---|---|---|---|
01301008 |
post | |||
The results for mean, geometric mean, quadratic mean, median, mode, min, max, sum, count, variance, population variance, standard deviation, total deviation, total covariance, total correlation, deviation matrix, covariance matrix, Pearson's correlation matrix, Spearman's correlation matrix, and Kendall's correlation matrix will be available. In the case of Linear Regression learning, it is recommended to have a correlation value between the predictor variable and objective variable of >0.7. |
||||
01301009 |
post | |||
The following results will be returned: mean, geometric mean, quadratic mean, median, mode, min, max, sum, count, variance, population variance, standard deviation, logarithm, logarithmic difference, ratio, logit, moving average, moving median, white noise, n-order differences, autocovariance, autocorrelation, partial autocovariance, and partial autocorrelation. (moveInterval: Calculation period for Moving Average or Moving Median; convertMode: Conversion Mode (0: original, 1: diff, 2: log, 3: log diff, 4: ratio, 5: logit); smoothMode: Smoothing Mode (0: none, 1: moving average, 2: moving median); diffOrder: Difference floors for Calculation) |
||||