2. Problem Statement
Agricultural practices can be improved in a number of ways. There are so
many things that farmers do manually that, with the advent of new
technology, can be improved upon to alleviate the burdens of the average
farmer. One of such things is being able to know, or at the very least, have an
accurate estimation of their crop’s yield before the harvest date. This would
help them make plans ahead in terms of storage space allocations and
expected profit. Before the advent of artificial intelligence, this prediction
has been carried out manually by the farmers. But as a result of human
shortcomings, they are only able to predict based on the crop’s performance
across a narrow range of years.
3. AIM & OBJECTIVES
However, with artificial intelligence techniques which can make comparison on datasets
spanning up to hundreds of years, this burden can be alleviated and more accurate
predictions can be made.
With the aid of machine learning, a subset of artificial intelligence, a model can be built and
trained to analyse similarities and relationships among various datasets on the same crop
across multiple years. The machine learning model will then be able to use the knowledge
that it has gained from these comparisons to be able to predict,with a high level of
competence, how future scenarios would play out.
It is with the implementation of machine learning, that we aim to develop an AI program that
can accurately predict the yield of a particular crop. .
To achieve this goal, we collected a large dataset of yield data for a specific crop under
various climate conditions (such as temperature and rainfall). We then preprocessed and
cleaned this data to prepare it for analysis. Using machine learning techniques, we trained a
model on this data to recognize patterns that were indicative of high or low yields. The
trained model was then tested on new, unseen data to evaluate its performance.
4. AIM & OBJECTIVES
Reiterating our problem statement, predicting crop yields accurately is
essential for optimizing production and improving the efficiency of
agricultural practices. However, given the delicate interplay of numerous
elements, such as weather conditions, crop management techniques, and
genotype, that can affect crop yields, this can be a difficult undertaking. As a
result, methods that can help with agricultural yield forecast are needed.
For this reason, we set out to create an AI software that can forecast the
yield of a specific crop using information on the crop's yields in previous years
and under varied climatic conditions. The goal was to help farmers and other
agriculture industry players optimize crop output and boost the effectiveness
of agricultural techniques by giving them useful insights.
5. Tools and Methods
Data collection: For a particular crop grown in a variety of climates, we
gathered a sizable dataset of yield data.
Upon data collection, we preprocessed and cleaned the data to make sure the
data was ready for analysis. Once we had gotten this data ready, we trained a
model on the preprocessed data using machine learning techniques. The
model developed the ability to spot data patterns that indicated high or low
yields.
We then evaluated the model using a variety of metrics, such as mean
absolute error and mean squared error to get an assessment of the model's
performance.
6. Tools
NumPy
Pandas
scikit-learn:
Seaborn
Matplotlib
In addition to the frameworks, algorithms which were also implemented in our program are:
OneHotEncoder
MinMaxScaler
train_test_split
The machine learning algorithm eventually used for our machine learning model is the:
Decision Tree Regressor Model, which is a type of machine learning model that can be used for
regression tasks (predicting continuous numeric values), They work by creating a tree-like model
of decisions based on the features of the data, with the goal of predicting the target value (in
this case, Crop Yield) as accurately as possible.
7. Recommendations & Conclusion
F. RECOMMENDATIONS
Based on the limitations of this project, we recommend the following:
High Quality Data collection
Higher Model complexity: To capture the full complexity of the factors that
affect crop yields, it may be necessary to use more complex models. However,
care should be taken to avoid overfitting, which can lead to poor performance
on new, unseen data.
Additional data to account for External factors
Minimized Human error
Testing and validation: It is important to thoroughly test and validate the
model on new, unseen data to ensure that it is accurate and reliable. This may
require collecting additional data for this purpose or using techniques such as
cross-validation.
8. Summary/ Conclusion
In this study, we demonstrated an AI software that forecasts a crop's yield based on
information about the crop's yields in previous years and under various climatic conditions.
We gathered a sizable dataset of yield data and trained a model on it using machine learning
techniques. The algorithm developed the ability to spot patterns in the data that indicated
high or low yields and then used these patterns to forecast outcomes for fresh, unforeseen
data.
With a mean absolute error of about 3% and a mean squared error of about 4%, we discovered
that the model was capable of producing reliable predictions. However, the availability and
caliber of the data, the generalizability of the model, and the model's complexity were some
of the project's drawbacks. We also highlighted human error and outside variables that may
have an impact on crop yields as potential causes of model inaccuracy.
In conclusion, this experiment shows how AI has the potential to be an effective tool for
predicting crop yields. But there is still potential for development in terms of the model's
accuracy and dependability. It might be possible to use more complex models, include more
information on outside influences, and improve quality control procedures in the future to
solve some of the constraints of this study. Additionally, it may be useful to explore the use of
AI in other aspects of agriculture, such as optimizing crop management practices or
predicting the impact of climate change on crop yields.