The problem Crop Yield-er solves

The modern world has come to an era of technological revolutions in every field. But regardless of this, farmers today face many challenges in meeting the growing demand for high-quality and high-yielding crops that are able to upkeep with the changing customer preferences. They must keep up with the new advances in technology, market trends, as well as shifting customer preferences related to their diets. In addition to this, they have to address issues like climate change, water scarcity, & soil degradation. Since all of these needs cannot be met by traditional agricultural practices, it is critical to modernize existing processes & use the data acquired over time to their benefit.

---

Once the perfect harmony is achieved and variables can be predicted better by the use of our AI model, it shall become easier for any farmer to implement better practices and methods that meet the requirements of contemporary farming and agricultural practices. This model can be succesfully used to derive and predict the yield by taking in available parameters such as rainfall, average temperature and amount of pesticides used during the crop period. We provide the functionality to calculate and predict yield with a desirable accuracy

---

• Our project aims at being an aid to farmers so that they can conquer the often humongous targets of contemporary agriculture. It helps the farmers in identifying the best farming practices by taking complete advantage of the conditions.
• It also gives them an idea of how to modify the current methods to increase the yield without having to increase work or labour
• Since it predicts the crop yield, we can depend on this model to control the Demand-chain management so that there is no supply-demand fluctuations in the market.
• This helps not only the farmers but also the common people who are greatly and directly impacted by the yield and its prices, which are bound to increase sharply in case of damage to crop or unexpected yield

Challenges we ran into

We faced an array of challenges during the completion of this project. These problems were tackled head-on by my team as we progressed the model to its completion for a noble cause.

---

1. Our initial dataset, formatted in JSON, proved to be cumbersome due to its heavy size and difficulties in downloading. We really put our efforts into working with the referenced dataset but since there is a time constraint to the competition we had to proceed with a dataset from Kaggle of our chosing.

• Not only we overcame it by using a more efficient CSV formatted Kaggle dataset but the comparision and training of the model became a lot easier once we shifted which streamlined the process.

2. When building and training our machine learning model, we encountered compatibility issues with the TensorFlow on MacOS, hindering its installation process.

• As a solution, we leveraged the Google Colab notebook to make things easier.

3. The polynomial regression model, with a degree of 3, initially encountered several type and value errors that required extensive time and effort to resolve.

• To solve these errors we had to extensively trace back to a point of correctness and alter the further progressions to change the outputs and produce desirable outcomes.

---

Despite these challenges, we were able to overcome them and achieve the required results.