DataFlow
Machine Learning Data Modeling Software
Machine learning algorithm data modeling software developed for model creation and training based on field-collected metrology data. DataFlow can be used in various measurement-based environmental fields, such as plant inflow and treatment water quality prediction, and membrane contamination analysis.
Easy to Start
DataFlow allows you to view and edit data in a spreadsheet and analyze data easily with simple settings.
- Import data from Excel (xls, xlsx) or text (csv) files
- Simple model training by providing training algorithm parameter default values
- Saving/loading trained model and prediction results
- Optimizing training settings defaults per model
- Modifiable algorithm parameters according to user needs
- Data pre-processing function (items with no data change among input/output items, outlier removal, etc.)
- Training and test data selection function (sequential, random, specified)
- Simultaneous training of multiple models and summary of training results
- Menus and instructions are available in both English and Korean.
- Online updates provided
Optimize Input Selection
DataFlow selects the optimal model by comparing the model training results according to various input combinations of the training model.
- Specify a range for the number of entries in the model
- Complete modeling for each combination of input items
- Results displayed in order of best performance
- Extracting variables with many high-performing models
- Copy and export analysis results
Basic Statistical Analysis
DataFlow provides univariate statistical calculations such as mean, variance, and standard deviation of user data in Excel format, and multivariate data analysis functions such as principal component analysis and correlation analysis. DataFlow can also perform pre-analysis to help you select inputs for your prediction/classification model.
- Basic statistics (mean, variance, standard deviation, maximum, minimum, median, etc.)
- Addition of Principal Component Analysis (PCA) and Score variable
- Correlation Analysis (CA)
- Add data transformation (Log, Power, Lead, Lag, etc.) variables
Various Analysis Functions
Various algorithms are provided to be used for modeling purposes. The settings of each algorithm use optimal values derived from various cases.
- PLS(Partial Least Squares)
- SVM(Support Vector Machine)
- MLP(Multi-Layer Perceptron)
- MLR(Multivariate Linear Regression)
- RANSAC(Random Sample Consensus)
Regression Analysis
- PLS-DA(PLS – Discriminant Analysis)
- DT(Decision Tree)
- Neural Network
- KNN(K-Nearest Neighbor)
- Kernel-SVM
Classification Analysis
- K-Means
- Balanced K-Means
Clustering Analysis
