INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

Development of a Smart Agricultural Marketplace with Machine

Learning-Based Price Forecasting

Ramaraj R

, Santhosh K

This paper presents a Smart Agricultural Marketplace integrated with machine learning–based price forecasting

to assist farmers in making informed selling decisions. The system predicts commodity prices using historical

agricultural market data and compares multiple regression models to identify the most effective predictor.

Linear Regression and Random Forest algorithms were trained and evaluated using realworld agricultural market

datasets. Experimental evaluation shows that the Random Forest model achieves superior performance,

obtaining an R² score of 0.9576 with significantly lower MAE and RMSE values compared to Linear Regression.

The results demonstrate that machine learning–driven price forecasting can provide reliable decision support and

reduce farmers’ dependence on intermediaries.

The dataset used in this study consists of 836,977 agricultural market records collected from publicly available

or incomplete entries were removed to ensure data integrity.

Agriculture continues to be one of the most significant sectors worldwide, providing food security and

employment to millions. However, traditional trading practices often create barriers that prevent farmers from

receiving fair prices for their produce. The major challenges include unpredictable price variations, poor access

to real-time market data, and dependence on intermediaries who reduce profit margins.

As a result, farmers frequently sell their goods at undervalued prices, while consumers pay inflated rates. The

introduction of datadriven technologies provides an opportunity to eliminate these inefficiencies by building

transparent and intelligent trading systems.

In recent years, Machine Learning (ML) and Artificial Intelligence (AI) have proven to be effective tools in

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

but often lack dynamic forecasting capabilities. These systems depend on static or manually updated prices,

which do not accurately reflect market trends.

To address this limitation, the proposed project introduces a Smart Agricultural Marketplace with an embedded

machine learning–based forecasting model that predicts future crop prices using historical market data, regional

information, and temporal price patterns.

The proposed architecture consists of several core modules: data collection, preprocessing, feature selection,

machine learning-based price forecasting, and a web-based trading interface. The collected datasets undergo

cleaning and normalization to ensure consistency and reliability.

Algorithms like Random Forest and Linear Regression are trained and evaluated to identify the most effective

model for price prediction. The system then integrates the forecasting module with a web application, allowing

but also strengthens the agricultural economy by encouraging sustainable and informed decisionmaking.

In recent years, machine learning (ML) and artificial intelligence (AI) have played a crucial role in transforming

agricultural systems through predictive analytics, automation, and intelligent decision support. Various

researchers have developed data-driven models to forecast crop prices, estimate yields, and improve transparency

in agri-trading systems. The following studies illustrate the major contributions and advancements in this area.

R. Kumar et al. [1] introduced a crop price prediction framework using Linear Regression and Random Forest

models. The system used data from local agricultural markets and weather conditions to estimate price

fluctuations for major crops such as rice and maize.

Their results demonstrated that Random Forest performed better in handling non-linear data, while linear models

authors emphasized the importance of including temporal factors and rainfall patterns for price prediction in

perishable commodities.

S. Reddy et al. [3] developed an IoT-integrated smart farming system capable of collecting real-time soil

moisture, temperature, and humidity data to predict both crop yield and price. Using Gradient Boosting for

prediction, their model achieved demonstrated improved predictive performance in forecasting short-term

market changes.

The system was extended with a mobile app interface that allowed farmers to view forecasted prices and

recommended the best selling periods, improved decision-support capability for farmers.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

directly with buyers, bypassing intermediaries. The platform incorporated a K-Nearest Neighbors (KNN) model

for localized price prediction and used blockchain for ensuring transparency in transactions.

Their results indicated that predictive pricing increased trust among users and reduced the exploitation of farmers

in the trading process. The study suggested that ML integration in e-marketplaces can substantially improve fair-

trade systems in rural economies.

P. Banerjee et al. [5] proposed a cloud-based machine learning system for crop price forecasting using Recurrent

Neural Networks (RNNs). Their dataset included five years of agricultural price records and weather data. The

study reported that RNNs outperformed traditional regression models in detecting seasonal trends and long-term

dependencies in agricultural datasets. The system demonstrated scalability by providing near real-time updates

for different regions, helping farmers plan their sowing and harvesting strategies efficiently.

Networks (CNNs) for analyzing satellite images with LSTM networks for time-series prediction. The fusion of

spatial and temporal data provided a comprehensive understanding of crop behavior and market trends,

improving overall prediction accuracy and enabling data-driven decision-making for farmers.

S. Mehta and B. Thomas [8] developed a regional agricultural price forecasting model using Bayesian Ridge

Regression combined with socio-economic indicators such as market demand, transportation costs, and fertilizer

usage. The study showed that including these external economic parameters improved prediction performance,

indicating that real-world agricultural prices are influenced by multiple interacting variables beyond crop and

weather data alone.

N. Ali et al. [9] implemented a deep reinforcement learning framework that dynamically adapted to market trends

for price prediction. The system learned from historical transactions to continuously adjust model parameters.

users to negotiate directly and perform secure online transactions. Evaluation metrics such as RMSE, MAE, and

R² demonstrated improved predictive performance, and user feedback suggested enhanced satisfaction and trust

among both farmers and buyers.

Overall, the literature demonstrates that machine learning-based forecasting systems are highly effective in

reducing uncertainty and improving profitability in agriculture. However, while numerous studies focus on

improving prediction accuracy, comparatively fewer works integrate forecasting models directly into digital

trading platforms, highlighting a research gap in combining predictive analytics with real-time agricultural

marketplaces.

The present research addresses this gap by developing a Smart Agricultural Marketplace that combines accurate

ML-based price prediction with a direct trading ecosystem, ensuring both transparency and accessibility for

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

trading while providing farmers with accurate price forecasts derived from historical market data.

The system follows a structured workflow:

1. Data Collection and Preprocessing – Historical agricultural market price data is collected from publicly

available agricultural datasets. The data is cleaned, validated, and normalized to remove inconsistencies and

ensure reliability.

2. Feature Extraction – Relevant attributes such as commodity type, state, district, market, minimum price,

maximum price, and temporal features (year, month, day) are extracted and prepared for model training.

3. Model Development – Machine learning models are trained on historical price records to learn patterns and

relationships influencing crop prices.

4. Model Evaluation – Models are evaluated using regression performance metrics to assess prediction

1. An ensemble learning algorithm that constructs multiple decision trees.

2. Captures nonlinear relationships in agricultural price trends.

3. Reduces overfitting through averaging.

4. Provides reliable predictions for structured tabular datasets.

1. Used as a reference model for performance comparison.

2. Provides interpretability and fast computation.

3. Helps quantify performance improvement achieved by the ensemble approach.

1. Data Preprocessing – Handling missing values, converting numeric fields, and encoding categorical

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue II, February 2026

(IJRASET), 2026.

3. Nitesh Singh & Ritu Sindhu, “Crop Price Prediction Using Machine Learning,” Journal of Electrical

Systems, Vol. 20, No. 7s, 2024.