Getting Started - RodentDataAnalytics/mwm-ml-gen GitHub Wiki
This section provides an overview of the mwm-ml-gen functionalities. For in-depth explanations please choose your topic from the sidebar on the right.
Contents
Main Menu
The main menu contains all the necessary tools for completing a behavioural analysis in a Morris Water Maze experimentation. The different steps of the analysis are arranged into distinctive panels starting from creating a new project or loading an existing one to generating various results.
During the analysis various files are generating which are arranged inside the specific folder the user has selected for his project; these files includes MATLAB MAT-files (.mat), which can be opened in MATLAB, comma separated values files (.csv), text files (.txt) and image files saved into various format such as JPEGs (.jpg) and PNG (.png). The numerical results are always generated inside CSV and TXT files and the graphs are exported into the user's defined image format.
New Project
When a new project has been specified the extracted data from Ethovision can be loaded for a pre-processing stage. In this stage the user can select with which animals he wants to work with, arrange them into groups and test his experimental properties by plotting their trajectories.
Segmentation
In the segmentation stage the animal trajectories are split into overlapping segments with length and percentage of overlap specified from the user. In this stage 8 features for the whole trajectories and the segments are also computed.
Labelling
In the labelling stage the user is provided with a graphical interface allowing him to visualise the trajectories or the segments and put appropriate labels to a portion of them. There is also the ability of exporting the non-labelled or the labelled segments to JPEG-image files.
Classification
The classification procedure has been updated to produce more accurate results. The user has the ability to use either the default settings or specify his own. Every step of the classification procedure can be manually configured.
Results
Various results in graphical and numerical way can be generated by the program such as: common metrics (latency, speed, path length), number or percentage of segments falling under each strategy, the probabilities for the animals to switch between the different strategies, the number of transactions between strategies, etc. Raw classification results can also been exported into CSV files for custom analysis.
