The IoT is one of the emerging sectors of society. As the use of IoT has increased the data from the sensors have also increased at a higher rate and a better and best computation is required. The computation models have to be analyzed by ML algorithms. These algorithms cannot be directly understood by a viewer and he/she expects the data to be more in visualizable format. For this purpose, we have created a frontend dashboard using the Atoti library. Atoti library in python is used to provide the user interface to view data as per their wish and the analyzed data using pandas, NumPy, and other libraries facilitated in better understanding of the plots when plotted against a different model of testing. The labeling of the data is taken care of by atoti itself and separate labeling isn't necessary. The Pycaret is used for the auto-ml part. By the end of this, we see that the ML models Accuracy and other parameters can be easily visualized by the user and he can change the format of his/her needs. This project is done by team VIT/OW/115 Members contributing to this project M Akshara, Sri Balaji J, R S Vimal.
Data Labeling and OLAP data visualization
OLAP Data Visualization helps create an interactive session for the users to dice, slice, filter, and carry operations to smoothly select a plotting framework. This helps users to have the working in a friendly manner and pleasingly view the plots. The data labeling helps users understand the data plot in a better format than the data that lack labeling. Atoti is one of the budding platforms to plot and visualize data and gives the user the option to visualize the data on his/her own with the required plots. Atoti's project implementation is limited to date. So this project will help users understand the ease of creating dashboards to visualize the data that they want.
Exploratory Data Analysis
Exploring data using analysis is one of the important criteria when dealing with data sets. The analysis of the data involves sorting the data into a more interpretable form for the machine to analyze using different models. This sorting of data can be done using the below libraries. The analysis stage begins by checking the row values and making the values in a single variable format throughout the column. Scikit-learn library can be used to individually model the data for example plotting decision trees, linear regression, etc. The panda's library provides a framework to view the column and row size and helps in the conversion of data frames to CSV and vice-versa. The number helps for checking null values or NaN values and replacing them with zero or a user-defined format. *Atoti is used for visualizing purposes* RECOMMENDATION: Use of seaborn for visualization of the models and use of matplotlib can also be used for data viewing purposes.
Pycaret library in python helps to understand the relationship of different models with the data set by using accuracy, AUC, recall, Precision, F1, TT, etc. This helps the user understand which model performs better and faster while handling that particular dataset. Pycaret has a few built-in commands which facilitate easy prediction of data using different inherited conditions. The auto-ml starter is one such package that helps in the easy usage of deploying techniques for predictive analysis. RECOMMENDATION: Auto-PyTorch and auto-sklearn are two other auto-ml tools that facilitate easier prediction.