Random Forest Classification Algorithm Research Articles

BridgeInsight: An asset profiler for penetration testing in a heterogeneous maritime bridge environment

A maritime bridge environment is a heterogeneous ecosystem of complex systems for various operations. As part of new requirements set by the International Association of Classification Societies, ship operators must now maintain an asset inventory aboard vessels specifically to improve their cyber security. This paper discusses the development of a ship-specific asset profiler that will not only identify and record the devices present automatically but also provide an in-depth analysis of their properties and characteristics in an intelligent and user-friendly manner. As cyberattacks increase in the maritime industry, proper testing of ship systems is essential, to ensure vessels remain secure and the risk of a cyberattack is minimized. An asset profiler for the bridge environment would serve as a tool for profiling the devices, helping personnel make faster and well-informed decisions, and could be a component of a wider audit framework. This paper presents a ship bridge profiler (i.e., BridgeInsight) used to identify all devices on the bridge of a vessel automatically and which provides information on them using a generated PDF report that consists of graphs and charts. To do this, it uses the Random Forest classifier algorithm, and the information it provides will enable the auditor or pen tester to perform manual testing or automate audits, while also providing comprehensive information that engineers and mariners can use to comply with regulations. Highlights As part of new requirements set by the International Association of Classification Societies, ship operators must now maintain asset inventory aboard vessels specifically to improve their cyber security. This paper presents a ship bridge profiler (i.e., BridgeInsight) used to identify all devices on the bridge of a vessel automatically. We envision automated asset detection and classification to have even more benefit in future cyber security work, as penetration testing.

Detection and Diagnostic with Random Forest Classifier (RFC) to Improve the Maintenance Management System in Steam Boiler of Power Plant

Industrial internet of things (IIoT), digital twin, and connected devices can continue to use smart equipment and improve access to data. While the data collected by sensors has been an invaluable asset to companies, the ability to understand and use this data to drive new insights. The development of Condition Monitoring (CM) technology and Computerized Maintenance Management System (CMMS) in power generation systems provides a validated set of operation and maintenance data with abundant event data. Maintenance decision-making is primarily based on equipment reliability and performance-based features for diagnosing equipment failure. The most critical asset and often reduces the reliability and availability of a Coal Fired Steam Power Plant (CFSPP) with the most frequency of disturbances is the steam boiler. As a departure from the idea of creating a digital twin, this article will focus on analyzing equipment health conditions and finding causes of failure of the tools, utilizing data for diagnostic purposes. Real-case used in this research are steam boilers, which are important assets in power plant generation. The online and Failure Mode and Effect Analysis (FMEA) module data will be combined to realize the concept of anomaly diagnosis which is driven by hybrid data. Hoping that accurate diagnosis result with the Random Forest Classifier (RFC) Algorithm can be obtained and be used to analyze the causes of failure and decrease in equipment performance resulting by a decrease of energy efficiency performance. The analytical approaches are carried out to have the goal of generating detection models and diagnostic insights of event data based on operational data and FMEA.

Spatially Explicit Modeling of Anthropogenic Heat Intensity in Beijing Center Area: An Investigation of Driving Factors with Urban Spatial Forms.

The escalation of anthropogenic heat emissions poses a significant threat to the urban thermal environment as cities continue to develop. However, the impact of urban spatial form on anthropogenic heat flux (AHF) in different urban functional zones (UFZ) has received limited attention. In this study, we employed the energy inventory method and remotely sensed technology to estimate AHF in Beijing's central area and utilized the random forest algorithm for UFZ classification. Subsequently, linear fitting models were developed to analyze the relationship between AHF and urban spatial form indicators across diverse UFZ. The results show that the overall accuracy of the classification was determined to be 87.2%, with a Kappa coefficient of 0.8377, indicating a high level of agreement with the actual situation. The business/commercial zone exhibited the highest average AHF value of 33.13 W m-2 and the maximum AHF value of 338.07 W m-2 among the six land functional zones, indicating that business and commercial areas are the primary sources of anthropogenic heat emissions. The findings reveal substantial variations in the influence of urban spatial form on AHF across different UFZ. Consequently, distinct spatial form control requirements and tailored design strategies are essential for each UFZ. This research highlights the significance of considering urban spatial form in mitigating anthropogenic heat emissions and emphasizes the need for customized planning and renewal approaches in diverse UFZ.

Procjena šumskih resursa na temelju Sentinel-2 snimki – studija slučaja Derventa, BiH (Katastarska općina Brezici)

The management of forest resources is complicated due to the complete lack of maintenance and disorganization of the land administration and survey that are decades old. Modern, unconventional monitoring systems are used with the aim of improving the existing records systems and creating a clearer insight into the state of forest resources. This study provides an example of the use of one such system, Sentinel-2. Using the R programming language, the multispectral Sentinel-2 images were classified by the Random Forest classification algorithm. Following the completion of the classifications, the accuracy of the classification was evaluated using the error matrix and the Kappa value. An analysis of forest resources for one cadastral municipality was accomplished using classified rasters and data from the Real Estate Cadastre Database. Based on the data analysis, major changes are visible in terms of the abandonment of agricultural land and its conversion into a certain form of forest vegetation. Furthermore, based on these data, the study demonstrates changes that can be monitored in shorter time intervals. Sentinel-2 images can be used to determine forest expansion, based on the aforementioned analyses, resulting in a clearer and better representation of existing forest resources that are unknown due to outdated and unreliable land administration systems.

E-mail Fraud Detection

Spam issues have become worse on social media platforms and apps with the growth of IoT. To solve the problem, researchers have suggested several spam detection techniques. Spam rates are still high despite the use of anti-spam technologies and tactics, especially given the ubiquity of rogue e-mails that lead to dangerous websites. By using up memory or storage space, spam e-mails may cause servers to run slowly. One of the most essential methods for identifying and eliminating spam is filtering e-mails. To this end, various deep learning and machine learning technologies have been used, including Naive Bayes, decision trees, SVM, and random forest. E-mail and Internet of Things spam filters use various machine learning approaches and systems are categorized in this research. Additionally, as more people use mobile devices and SMS services become more affordable, the issue of spam SMS messages is spreading worldwide. This study suggests using a variety of machine learning approaches to detect and get rid of spam as a solution to this problem. According to the trial findings, the TF-IDF with Random Forest classification algorithm outperformed the other examined algorithms in accuracy %. It is only possible to gauge performance on accuracy since the dataset is imbalanced. Therefore, the algorithms must have good precision, recall, and F-measure.

Machine learning-based prediction for self-harm and suicide attempts in adolescents

This study aimed to use machine learning (ML) models to predict the risk of self-harm and suicide attempts in adolescents. We conducted secondary analysis of cross-sectional data from the Longitudinal Study of Australian Children dataset. Several key variables at the age of 14–15 years were used to predict self-harm or suicide attempt at 16-17 years. Random forest classification models were used to select the optimal subset of predictors and subsequently make predictions. Among 2809 participants, 296 (10.54%) reported an act of self-harm and 145 (5.16%) reported attempting suicide at least once in the past 12 months. The area under the receiver operating curve was fair for self-harm (0.7397) and suicide attempt (0.7220), which outperformed the prediction strategy solely based on prior suicide or self-harm attempt (AUC: 0.6). The most important factors identified were similar, and included depressed feelings, strengths and difficulties questionnaire scores, perceptions of self, and school- and parent-related factors. The random forest classification algorithm, an ML technique, can effectively select the optimal subset of predictors from hundreds of variables to forecast the risks of suicide and self-harm among adolescents. Further research is needed to validate the utility and scalability of ML techniques in mental health research.

A simple Node-RED implementation for digital twins in the area of manufacturing

Interest in digital twins continues to strengthen with technological advancements in Industrial IoT. A digital twin is a virtual representation that models a physical object and effectively provides a two-way interaction with the real system. Digital twin models can be set up to test or analyze industrial applications before deployment thereby improving the efficiency of industries. In this work, a Node-RED implementation for digital twins in the manufacturing sector is developed. Plastic injection molding is the chosen case study for the implementation of this digital twin. Node-RED is a platform that allows developers to quickly build Internet of Things applications using a simple web browser interface. The digital twin uses the Random Forest Classifier algorithm to do predictive maintenance tasks including classification of quality of products. An easy-to-use dashboard is developed to enable the user to interact with the digital twin. Important modules such as communication with the real environment, SMS, and email notifications are successfully implemented in the digital twin. The findings show that it is feasible to build a Node-RED digital twin. The flexibility of Node-RED makes it suitable for building architecture of varying complexity.

Analysis of the Customer Churn Prediction Project in the Hotel Industry Based on Text Mining and the Random Forest Algorithm

The ability of hotels to differentiate themselves from competitors and continue to operate profitably depends on their ability to retain their customers by building long-term and permanent customer relationships. Technological developments in recent years have made it possible for companies to predict their customers’ behavior by accessing their opinions faster and preventing them from churning. Managing customer churn prediction projects has become an important issue today, especially in the hotel industry. Therefore, this research seeks to analyze projects that predict the churn of hotel customers to provide a model to help hotel managers in this field. In this research, an approach based on text mining on customers’ comments in the Persian language is presented, which uses the random forest algorithm for classification that was considered the most effective method to solve this problem. In this model, to increase the efficiency of the proposed method in compare with existing works, the gravitational search algorithm was used to select the useful features, and the differential evolution algorithm was used to adjust the parameters of the classification method. The dataset of this research is the collected data from the customer database on social networks and hotels’ websites, especially the hotels on Kish Island in Iran. The results of this research showed that after the implementation of the preprocessing operations, the method of adjusting the parameters and removing the unimportant features, the model’s accuracy increased significantly. The precision, recall, F1, and accuracy criteria were 0.77, 0.76, 0.76, and 0.77, respectively.

Improving the accuracy of sentiment analysis using a linguistic rule-based feature selection method in tourism reviews

Sentiment Analysis technique involves extracting the relevant information from Unstructured User Reviews (UUR) dataset fetched from online and classifying them into appropriate positive and negative comments for making decisions. In UUR, data may be in noisy state, irrelevant features exist which creates high dimensional feature space. To design an effective sentiment learning model, users are required to extract the most relevant sentiment features from UUR. To overcome the issue, we proposed a Linguistic rule based feature selection method for extracting and selecting the sentiment features for Sentiment Analysis as it improves the predictive performance of classification algorithms. The proposed novel feature selection method involves identifying the various sentiment features in the review dataset by using filtering methods such as POS tags, n-grams. In the ensemble model, where the Random Forest classification algorithm is trained for textual sentiment classification, the chosen sentiment feature sets are used. Finally, we test our approach using the real-time review dataset that was collected from a multitude of sources, and the results demonstrate prediction accuracy that is superior to that of existing Sentiment analysis techniques.

Mapping lichen abundance in ice-free areas of Larsemann Hills, East Antarctica using remote sensing and lichen spectra

Lichen communities are known to be most resistant and adapted organisms to the extreme environments; however, their abundance is not well mapped. Extensive lichen surveys were conducted as part of the 39th Indian scientific expedition and in-situ spectra (350 nm–2500 nm) of lichens were collected in the Larsemann Hills, East Antarctica during austral summer of 2020. Lichen abundance mapping was carried out with the help of Sentinel-2 MSI L2 data and surveyed records along with in-situ spectra. We generated feature collections for lichen, snow, water, bare surface and trained a random forest (RF) classification algorithms implemented in GEE and generated multi-class outputs. We finally merged all non-lichen classes and produced binary pixels with a confidence value (between 0 and 100) depicting similarity of its spectral response to that of a lichen pixel. Total 92 lichen points, 20 bare rock points, 26 points of water and 74 snow points were used to generate the probabilistic lichen abundance map. Resubstitution accuracy of 97.31% was obtained with 10 number of RF trees. Validation was done with geotagged ground photographs having 232 lichens, 20 bare rocks, 22 water and 69 snow points and achieved test accuracy of 82.44%.

Major Determinants of Bank Profitability in India: A Machine Learning Approach

From reforms and fin-tech revolutions to macro-economic shocks, the Indian banking sector has witnessed rapid changes over the last two decades, which has significant implications for banks’ profitability. Viewing bank profitability from three different dimensions, Net Interest Margins (NIM), Return on Assets (RoA) and Return on Equity (RoE), this study has explored the key determinants with the help of machine learning algorithms. It has used a pooled data set of domestic and commercial banks covering 2005–2021. As a dependent variable, profitability by each measure (NIM, RoA and RoE) is reclassified into three categories, above average, average and below average, based on their quartiles. Twenty-one explanatory variables comprising bank-specific, macroeconomic and policy variables are chosen after due validation using feature selection methodology and multicollinearity check. The random forest (RF) classification algorithm is executed using the CARET package in R. The results obtained from feature selection are corroborated with the RF classification findings. The results are robust and give clear-cut visibility of unique and common factors influencing three profitability measures at varying levels. The classification estimates suggest that the bank-specific variables are major determinants of NIM, while macroeconomic and policy variables are the key determinants of RoA and RoE. Further, the results also suggest that the ratio of non-performing assets to total assets and business per employee are two such bank-specific determinants that play an important role in all three dimensions of profitability. Thus, recapitalization and automation will play an important role in bank profitability.

Blockchain and Classification of Mammograms and Histopathology Images in Breast Cancer Lesions

This work provides an overview of the research that has been done on using Computer-Aided Detection (CAD) for the diagnosis of breast cancer. The focus is on mammographic and histopathology images and the different techniques used for image pre-processing, segmentation, and classification. The accuracy of the different algorithms was evaluated on different datasets, including MIAS, IRMA, DDSM, and CBIS-DDSM, and the results showed that deep learning models such as Convolutional Neural Networks (CNNs) and Random Forest, along with Multi-Layer Perception and Nave Bayes, were effective in detecting and classifying breast cancer. The results of these studies show the potential of CAD in making the diagnosis of breast cancer easier and more accurate. In the segmentation stage, the U-Net model has been modified to perform better in terms of accuracy. The DDSM database has been found to have a higher accuracy percentage compared to the MIAS and CBIS-DDSM databases. This indicates that the modified U-Net model has performed well on the DDSM database in terms of accurately segmenting the images. The use of Computer Aided Diagnosis (CAD) has improved the accuracy of breast cancer diagnosis in mammography and histopathology images. The four-step process of preprocessing, segmentation, feature extraction, and classification has proven to be effective in detecting malignant and benign tumors. Different algorithms like Naive Bayes, Multilayer Perception, Random Forest, and Convolutional Neural Network (CNN) have been used with varying accuracy levels. The best results were obtained by using the modified U-Net model for segmentation and the Random Forest algorithm for classification, as they showed higher accuracy compared to other methods and databases. In this study it has been analyzed that the use of Computer-Aided Diagnosis (CAD) in mammography and histopathology images has shown promising results in the diagnosis of breast cancer. Different algorithms and techniques have been used to improve the accuracy of the classification of malignant and benign tumors. The MIAS database has shown the best results when using a Convolutional Neural Network approach. However, histopathology images have some limitations in comparison to radiologist images, but the use of CAD is still an important tool in cancer diagnosis.

Random Forest Optimization Using Particle Swarm Optimization for Diabetes Classification

Diabetes mellitus is a chronic degenerative disease caused by a lack of insulin production in the pancreas or the body's ability to use insulin less effectively. According to a report by the World Health Organization (WHO), 4% of the total deaths in the world are caused by diabetes. The International Diabetes Federation (IDF) notes that in 2013 there has been an increase in diabetes sufferers. Indonesia is the seventh place with the largest number of cases of diabetes mellitus. In this study, the method used to classify diabetes is using a random forest algorithm with Particle Swarm Optimization (PSO) optimization. This study resulted in an accuracy of the random forest classification algorithm of 78.2% and 82.1 using PSO optimization with an increase in value of 3.9%. It can be concluded that PSO optimization can provide a better increase in classification accuracy values when compared to the random forest algorithm without PSO optimization

Classification of Poisonous and Edible Mushrooms with Optimized Classification Algorithms

Within the scope of this study, it is aimed to classify the mushroom species consumed as a staple food. For this purpose, 8124 mushroom data with 22 different mushroom feature information were used. 5686 of these data were used for training and 2438 for testing. In the study, poisonous and edible mushroom species were classified by random forest, decision tree, and logistic regression classification methods. The parameters used in the random forest and decision tree classification algorithms used in the study were optimized with the GridSearchCV optimization method. With the random forest algorithm, the highest precision, recall, and F1 score values are 0.93, 0.98, and 0.95, respectively. When these values are examined on a class basis, the highest distinctiveness results were obtained in the poisonous class. In the edible class, the highest performance results were measured as 0.97, 0.92, and 0.95 for precision, recall, and F1 score values, respectively. With the decision Tree algorithm, the highest precision, recall, and F1 score values are 0.98, 0.98, and 0.92, respectively. The highest precision, recall, and F1 score values of the best poisonous class are 0.90, 0.98, and 0.92, respectively. The best performance results of the edible class were obtained with the highest precision, recall, and F1 score values of 0.98, 0.89, and 0.90, respectively. The average accuracy rate was 0.9028 with the Logistic Regression algorithm, and the precision, recall, and F1 score values of the poisonous class were obtained as 0.86, 0.97, and 0.91, respectively. Precision, recall, and F1 score values of the Edible class were obtained as 0.96, 0.83, and 0.89, respectively.

Development of New Machine Learning Based Algorithm for the Diagnosis of Obstructive Sleep Apnea from ECG Data

In this study, a machine learning algorithm is proposed to be used in the detection of Obstructive Sleep Apnea (OSA) from the analysis of single-channel ECG recordings. Eighteen ECG recordings from the PhysioNet Apnea-ECG dataset were used in the study. In the feature extraction stage, dynamic time warping and median frequency features were obtained from the coefficients obtained from different frequency bands of the ECG data by using the wavelet transform-based algorithm. In the classification phase, OSA patients and normal ECG recordings were classified using Random Forest (RF) and Long Short-Term Memory (LSTM) classifier algorithms. The performance of the classifiers was evaluated as 90% training and 10% testing. According to this evaluation, the accuracy of the RF classifier was 82.43% and the accuracy of the LSTM classifier was 77.60%. Considering the results obtained, it is thought that it may be possible to use the proposed features and classifier algorithms in OSA classification and maybe a different alternative to existing machine learning methods. The proposed method and the feature set used are promising because they can be implemented effectively thanks to low computing overhead.

A dataset of land surface water with a spatial resolution of 30 meters on the Qinghai-Tibet Plateau in 2022

The Tibetan Plateau is known as the Asian Water Tower. The distribution of surface water and its changes are closely related to global change, biodiversity and water-related ecosystems. Based on the collection of high-precision land surface water samples, we used the random forest classification algorithm in machine learning to extract land surface water information from Landsat series satellite images and produced a dataset of land surface water with a spatial resolution of 30 meters on the Qinghai-Tibet Plateau based on satellite remote sensing images in 2022. According to data quality assessment, the overall accuracy of the dataset is 92.9%, and the Kappa coefficient is 0.84. This dataset can provide foundational data support for water resource monitoring, ecosystem services, and global change research on the Qinghai-Tibet Plateau.

Influence of Urban Land Use Land Cover Changes on Land Surface Temperature in Dar es Salaam Metropolitan City, Tanzania: The Use of Geospatial Approach

Land use land cover (LULC) changes affect the planet's energy balance and region's climate. Land Surface Temperature (LST) is a vital indicator of this change. Studies in Dar es Salaam Metropolitan City have investigated LST and its relationships with building heights and densities, urban heat islands, spectral indices, and urban morphological determinants. The present study used cross-sectional profiles, chord diagrams, and simple linear regression models to examine the influence of LULC changes on the LST in Dar es Salaam Metropolitan City (DMC). LST was extracted from Landsat 5 TM and 8 OLI/TIRS images for 1995, 2009, and 2017. LULC was identified via the supervised random forest classification algorithm. Between 1995 and 2017, built-up areas rose by 8%, vegetation fell 7%, and bare soil 3%. As a result, the average LST rose by 3 °C. Built-up areas had the highest temperatures (24–26.5 °C), followed by bare soil (22–25.5 °C). The lowest temperatures (21–25 °C) were on vegetation and water. Built-up area positively correlated with LST, while vegetation, water bodies, and bare soil negatively correlated. The study results can assist local authorities in enforcing urban planning regulations, raising public awareness, and guiding policymakers in creating sustainable planning and management strategies for the future.
 Keywords: Dar es Salaam, Land use land cover, simple linear regression model, land surface temperature, chord diagrams

Effects of Temporal and Spatial Changes in Wetlands on Regional Carbon Storage in the Naoli River Basin, Sanjiang Plain, China

The Naoli River (NLR) Basin is a crucial distribution area for wetlands in China. Investigating the link between land use changes and carbon storage in this basin is of significant importance for protecting regional ecosystems and promoting the sustainable development of the social economy. This paper uses long-term Landsat satellite images provided on the GEE (Google Earth Engine) platform and the random forest classification algorithm to create spatial distribution maps of land use in the NLR Basin from 1993 to 2022. The study analyzes the dynamic changes in wetlands in the basin over the past 30 years and employs the InVEST (Integrated Valuation of Ecosystem Services and Tradeoffs) model to explore the temporal and spatial evolution characteristics of carbon storage. The results reveal that the wetland area of the NLR Basin showed a downward trend from 1993 to 2022, with a total decrease of 1507.18 hm2 over 30 years. During this period, the carbon storage in the NLR Basin decreased, with a cumulative loss of 1.98 × 107 t, mainly due to the continuous reductions in wetland and forest land. Additionally, the change in carbon storage in the basin has a strong spatial and temporal relationship with the changes in land use/cover area. The total carbon storage is positively associated with the areas of wetland, forest land, and water bodies. The conversion of wetlands into any other land type results in the reduction in carbon storage. These findings can improve our understanding of the spatial and temporal dynamics of wetlands in the NLR Basin over the past 30 years and enable us to analyze the relationship between land use changes and regional carbon storage. The results of this study have great significance for protecting the wetland ecology and regional carbon balance in the NLR Basin.

Construction of multidimensional features to identify tea plantations using multisource remote sensing data: A case study of Hangzhou city, China

China is the world's largest tea producer and trader. From 1995 to 2020, the plantation area and tea production increased by approximately 200 and 400%, respectively. Therefore, we must accurately identify the distribution of tea plantations to optimize production and promote sustainable development. However, most studies on the identification of tea plantations rely on local hyperspectral data processing, which have limitations related to a small scale, significant time consumption, and difficulty in integrating multisource images and multidimensional features to achieve the accuracy and optimization of the results. Based on the Google Earth Engine (GEE) platform, we integrated the spectral, synthetic aperture radar (SAR), terrain, and textural features to extract the seasonal phenological index using specific phenological information on tea plantations. Then, the random forest (RF) algorithm was applied to evaluate the importance of each feature in the initial feature set, followed by screening of the optimal feature combination and verification of the Jeffries–Matusita (JM) distance. Finally, the optimal feature combination was used as the input variable for the RF classification algorithm to identify the distribution of tea plantations in Hangzhou, China. The following outcomes were achieved: (1) compared to the processing of local hyperspectral data, the identification of tea plantations using the GEE had notable advantages, such as high automation, convenient algorithm adjustment, and rapid iteration of the results. (2) The user and producer accuracies of the classification of tea plantations using SAR and terrain and textural features improved from 89.3 and 85.9% to 95.8 and 91.7%, respectively, compared to only using spectral information. (3) The high-contribution feature combination identified by the importance analysis using the RF algorithm increased the JM distance to a high degree of separability of approximately 2, which is highly effective for the optimization of classification features. (4) In addition to the visible and near-infrared bands commonly used in spectral information, short-wave infrared and vegetation red-edge bands can play important roles in identifying tea plantations. (5) The importance of the same spectral or SAR information varied for different seasonal phenological indicators, indicating that the addition of phenological information was beneficial for the identification of tea plantations.

Early Crop Mapping Based on Sentinel-2 Time-Series Data and the Random Forest Algorithm

Early-season crop mapping and information extraction is essential for crop growth monitoring and yield prediction, and it facilitates agricultural management and rapid response to agricultural disasters. However, training classifiers by remote sensing classification features for early crop prediction can be challenging, as early-season mapping can only use remote sensing image data during part of the crop growth period. In order to overcome this limitation, this study takes the Sanjiang Plain as an example to investigate the earliest identification time of rice, maize and soybean based on Sentinel-2 time-series data and the random forest classification algorithm. Crop information extraction was then performed. Following the analysis of the remote sensing classification features by the random forest importance approach and the subsequent normalization, the optimal features greater than or equal to 0.5 have yielded quite results in early crop mapping, and their overall accuracy was the highest in early-season mapping. The overall accuracy was observed to improve by 5% for 10 to 20 days of delay. In addition, rice, maize, and soybean were mapped at the irrigation transplanting period (10 May), jointing stage (9 July) and flowering (29 July), with an overall accuracy of 90.4%, 90.0% and 90.9%, respectively. This study shows that features suitable for early crop classification can be selected by random forest importance analysis as well as the ability of remote sensing to extract crop acreage information within the reproductive period.