The Data Explorer by Ubidots is a collection of data analysis tools to help identify and analyze trends, anomalies, and correlations across variables in an App. With the Data Explorer you can:
- Historical data navigation: Analyze data across different time periods on the go.
- Correlation analysis: Compare two or more variables in the same chart, or across several charts
- Max/Min analysis: Graph variables along with their rolling means, max, min, or sums.
- Regression Analysis: Apply regression algorithms to create trend lines and insights.
- Anomaly Detection: Perform moving average analysis for anomaly detection
Table of Contents
- Getting started with the Data Explorer
- Navigating the Data Explorer
- Example Analysis
1. Getting started with the Data Explorer
Access the Data Explorer from "Data" –> "Analytics" of your Ubidots Admin account Navbar.
To get started, simply click on “Add variables”, then select the variable(s) you wish to include in the analysis:
After adding your variables, you'll be able to modify their properties using the vertical configuration menu on the right of the Explorer user interface. You can set the type, Y-axis range and aggregation method of each variable to be displayed.
2) Navigating the Data Explorer
The Data Explorer interface is made up of three components:
1. Time Frame Selector
Data Explorer analyzes data from the chosen Start Date with Span (which is the amount of minutes, hours or days to include after the start date) to generate insights.
In additional to select the Start Date and data Span you must select an aggregation method (count, min, max, etc) and a Sample Period (the sample time period of the entire Time Frame selected).
Aggregation Method Options
- Count: The amount of data points in each Sample Period
- First: The first data point of each Sample Period
- Last: The last data point of each Sample Period
- Max: The max value of each Sample Period
- Min: The min value of each Sample Period
- Sum: The sum of all values within each Sample Period
2. Data Configuration Menu
From the right-side of the Data Explorer user interface you are able to add and remove variables for your analysis.
Based on the data configuration and time frame, you are able to view real-time charts to analyze, understand, and digest data more easily.
3) Example Analysis
Adding a Moving Average
To create a moving average across your time frame, click on “Add Variable” in the configuration menu, then click on “Moving Average”:
Then, select the variable from which you'd like to compute the Moving Average:
A new variable with the Moving Average will be created.
Adding a Regression Trendline
To add a trendline, click on “Add Variable” in the configuration menu, then select “Regression":
Then, select the variable from which you'd like to compute the Regression; a polynomial regression trendline will be added your chart:
You can modify the properties of the regression, including the degree of the polynomial within the Configuration panel. We'll also display the polynomial coefficients to allow you to easily replicate this in external tools such as Excel:
The polynomial coefficients can be interpreted using the following canonical expression:
- Is the time relative to the first data point of the series being displayed
- Has the same time units as the current Sample Period in the date picker
- Ubidots computes the regression trend using the dots (data points) currently being displayed in the Data Explorer Chart. Ubidots does not compute the regression using the entirety of the variable's time frame.
- Ubidots doesn't use milliseconds as the time unit to compute the regression, as it would yield very large-magnitude coefficients. For this reason, the sample period is used.
The Cloud is the limit - what is in Ubidots Data Explorer Roadmap…
We see the evolution of Statistical and Machine Learning methods as an opportunity for IoT entrepreneurs to enhance their solutions and add more value to their customers. As a next step towards enriching our Analytics Data Explorer module, Ubidots is working on an machine learning mechanism for anomaly detection.
How would you like to apply Machine Learning to IoT? Feel free to drop us your comments and suggestions using our in-app chat from the Data Explorer user interface.
Other users also found helpful...