To train a model that can predict the variations of temperature and humidity conditions for more efficient management of resources.
We started by gathering poultry data from over 1000 poultry farmers in Ndeiya(a highland area in Kenya). We then had to calculate the statistical values of the data set(means, median, and mode) which then helped us to clean the dataset(i.e fill any gaps within the data set to avoid disparities) By Leveraging on Machine Learning models provided by Microsoft Azure Learning Studio, we had to columnize tha data. Considering that we took very different mesurents(for instance temperature, humidity, soil mositure, and frequency of disease occurrence), we had to specify what dataset we wanted to use in our model. As such, we divided the data set into columns and utilised only the temperature and humidity datasets. From the dataset we then had to split the dataset, using half of it to train the model and half of it to test the model's accuracy. We then used a multiclass neural network to constantly reinforce the model's ability to predict. Therafter, after running the model with the dataset, we evaluated the model and achieved an accuracy of 76%. We then repeated the process severally which required us to use more data in our training. Eventually we piloted the model to over 500 farmers.