Inquiry About Real-Time Water Quality Index Prediction Using ThingSpeak

I have been actively using ThingSpeak for visualizing real-time data from pH and TDS (Total Dissolved Solids) sensors in water quality monitoring. ThingSpeak has been instrumental in providing a user-friendly platform for data visualization, and it has greatly enhanced our monitoring capabilities.However, as our project advances, we are interested in taking a step further to predict the Water Quality Index (WQI) in real-time using data collected from the pH and TDS sensors. We have developed a predictive model to estimate the WQI based on the dynamic values of pH and TDS, and we are eager to integrate this model with ThingSpeak for real-time predictions.Our objective is to leverage the capabilities of ThingSpeak to not only visualize the data but also to utilize the predicted WQI for enhanced water quality monitoring. By doing so, we aim to provide a more comprehensive and timely assessment of water quality, enabling better decision-making and more proactive actions.I kindly request your guidance and expertise regarding the feasibility of deploying our predictive model within the ThingSpeak platform. We are particularly interested in understanding if it is possible to run the model externally through ThingSpeak, and if so, what would be the necessary steps and considerations to make this a reality.Your insights and recommendations would be of immense value to our project, and we are excited about the potential to enhance our water quality monitoring system through real-time predictive analytics.Thank you for your time and consideration. We look forward to your response and the opportunity to explore this exciting possibility further.

Answer 1

[nick]

Hi Kola,

I understand that you are facing an issue related to deploying your predictive model to ThingSpeak Cloud and require guidance in resolving the same.

The following article ,"Developing an IoT Analytics System with MATLAB, Machine Learning, and ThingSpeak", demonstrates the preferred approach to train the model on a desktop and embed the trained model parameters in the MATLAB code that is operationalized on ThingSpeak :

https://www.mathworks.com/company/newsletters/articles/developing-an-iot-analytics-system-with-matlab-machine-learning-and-thingspeak.html

Another, less preferred, approach is to upload the trained model (the .mat file) to Dropbox by using :

https://www.mathworks.com/matlabcentral/fileexchange/59673

and have the MatLAB Code that ran in ThingSpeak to downlaod the model at runtime using :

https://www.mathworks.com/matlabcentral/fileexchange/67833

Next you can use the trained model in your code to make a prediction with your real time data in ThingSpeak. The new data is fetched from ThingSpeak field using 'thingSpeakRead'.

Kindly refer the attached MATLAB answer for more information on deploying trained machine learning model to ThingSpeak:

https://www.mathworks.com/matlabcentral/answers/516865

Hope this helps,

Regards,

Neelanshu

[kola veera venkata bhavani]

Just one more question can we use models like LSTM instead of NARX neural network algorithm to deploy in th ThingSpeak?

[Christopher Stapels]

Yes it is possible to use LSTM for the two methods above, but there are some limitations in each method.