<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Abhinav Baruah on Medium]]></title>
        <description><![CDATA[Stories by Abhinav Baruah on Medium]]></description>
        <link>https://medium.com/@baruah1993?source=rss-a3bb01a6f99a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*FI7DPadYOet4rH-cf9YsTg.jpeg</url>
            <title>Stories by Abhinav Baruah on Medium</title>
            <link>https://medium.com/@baruah1993?source=rss-a3bb01a6f99a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 04:00:01 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@baruah1993/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Vulnerability Mapping of Undernutrition using Machine Learning]]></title>
            <link>https://medium.com/analytics-vidhya/vulnerability-mapping-of-undernutrition-using-machine-learning-aa27907b6f0d?source=rss-a3bb01a6f99a------2</link>
            <guid isPermaLink="false">https://medium.com/p/aa27907b6f0d</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[nutrition]]></category>
            <category><![CDATA[public-health]]></category>
            <category><![CDATA[food]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Abhinav Baruah]]></dc:creator>
            <pubDate>Tue, 01 Sep 2020 20:07:12 GMT</pubDate>
            <atom:updated>2020-09-02T13:24:21.661Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*uBHMoh8aIukex5Ld.jpg" /><figcaption><a href="https://blog.forumias.com/indias-malnutrition-challenge/">Source</a></figcaption></figure><h3><strong>Introduction</strong></h3><p>The Global Nutrition Report 2018 defines undernutrition as:</p><p>“<em>lack of proper nutrition, caused by not having enough food, not eating enough food containing substances necessary for growth and health, and other direct and indirect causes.</em>”</p><p>With an alarming rank of 102nd out of 117 qualifying countries in the Global Hunger Index 2019, India suffers from a level of hunger that is identified as serious. Despite having enough food production, India suffers from a ‘hidden hunger’ due to insufficiency in fulfilling the nutrient requirement of its population. Thus prior identification of threat to nutrition security is necessary to formulate food security policies by the government. In order to address the issue, we would apply machine learning techniques to predict vulnerability of Indian districts to undernutrition among newborn children. A birth weight less than 2.5 kg is conventionally considered ‘low’ in most countries. A low birth weight (LBW) is not only associated with increased risk of child mortality but also affects growth and development during adolescence and triggers adult diseases such as coronary artery disease and diabetes. [​1​] Thus percentage of LBW prevalent in a district is a crucial indicator of undernutrition among newborn children.</p><p>The prime objective is to classify LBW vulnerability of a district into “Low”and “High” level based on socio-economic and health parameters by building our own ensemble ML classifier. Hence visualize the results by generating visual heat maps of vulnerability levels of all districts in a state.</p><h3><strong>Methodology</strong></h3><p>We have employed both supervised and unsupervised learning. Clustering techniques have been used to find out optimal clusters in the data and classification techniques have been finally used to classify into labels. Regression (Linear and Logistic) have also been used in classification. Apart from this, outlier detection algorithms have been used to detect and remove outliers. Moreover preliminary exploratory data analysis (EDA) has also been performed on the dataset,before designing the model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/668/1*kQasNok8WULRStgS0LCGcg.png" /></figure><p>We have used the <strong>Annual Health Survey(AHS) of India</strong> data for the year 2012–13 to train our model. ​The survey was conducted in Empowered Action Group (EAG) states Uttarakhand, Rajasthan, Uttar Pradesh, Bihar, Jharkhand, Odisha, Chhattisgarh &amp; Madhya Pradesh and Assam. These nine states are the high focus States in view of their relatively higher fertility and mortality. It is a representative sample of about 21 million population and 4.32 million households and covers 20k+ sample units which is spread across rural and urban areas of these nine states.</p><p><strong>Number of variables:</strong> ​41</p><p><strong>Dependent variable:</strong> ​<em>Percentage Children with birth weight less than 2.5 kg</em></p><p><strong>Independent variables:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/791/1*9JhMSCmuM2D6IaFeHT2MpQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/740/1*BrcGp4lj3a6F8wzmLPMUPA.png" /></figure><p><strong>Techniques:</strong></p><p><strong>Local Outlier Factor:</strong> It is a density based outlier detection algorithm.This algorithm evaluates each datapoint’s uniqueness on the basis of distance from the k-nearest neighbours. This method is based on the local density and compares the datapoint’s local density with the nearest neighbour density. [​2​] A datapoint is considered an outlier if its density is much low compared to its neighbour. The local outlier factor (LOF ) of a datapoint p is given by [​2​]:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/339/1*oT-J9hHQ1oakYqq_fbWXZw.png" /></figure><p>where local reachability density (lrd) of the object p is given by[​2​]:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/341/1*7rGV4DSnyuMo-26y-qzNVw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/661/1*DHtxIv54DRuAplc8VIj4og.png" /></figure><p><strong>Clustering techniques:</strong> Clustering was done on the unsupervised unlabeled data to find optimal clusters forming naturally within the data. K-Means and Hierarchical Clustering were performed for this purpose. The silhouette coefficient and J vs k plot (Elbow method) were used to determine the optimal clusters k. Based on the results of clustering, two levels were created Low ( less than median) and High (greater than median).</p><p><strong>Classification Algorithms:</strong></p><ol><li>Linear Regression</li><li>Logistic Regression</li><li>K- Nearest Neighbor (KNN)</li><li>Decision Tree</li><li>Random Forest</li><li>Adaptive Boosting</li></ol><p><strong>Dimensionality Reduction Techniques : </strong>We also used <strong><em>Linear Discriminant Analysis (LDA)</em></strong> dimensionality reduction technique to carry out classification. First from the independent variables, test and training sets were separated in 1:2 ratio.These training and test sets were then scaled based on statistical parameters of the training set . Then the dimensionality of the variables was reduced using LDA and the first component LDA 1 was selected to perform classification using KNN and Random Forest Classifiers.</p><p><strong>Ensemble Learning :</strong><br>By evaluating the results from all the classification techniques used above, we chose the top three classifiers based on the metrics score to design our own ensemble technique. The ensemble classifier takes the predictions from the top three classifiers as input and selects the best prediction based on majority voting criteria. It has been designed to achieve maximum accuracy by combining all the three classifiers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/636/1*GOmpQijPYNFE4IALkGrD6A.png" /></figure><p><strong>Model evaluation:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/731/1*rohcZsRh1daPYIK5mSuIWg.png" /></figure><p><strong>F1-score</strong> = (2 * Precision * Sensitivity) / (Precision + Sensitivity)</p><h3><strong>Results</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5NDZgbsBITQgUTOafX0P7A.png" /><figcaption>Figure: correlation matrix of independent variables</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/138/1*GojxC3t7QHDpoS4n3V6k5A.png" /><figcaption>Figure: Box-plot of dependent variable</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/677/1*pLeUHlgex83qqER57CNRgQ.png" /><figcaption>Figure: Histogram of dependent variable</figcaption></figure><p><strong>Clustering:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/999/1*YWDq7d9kFMe0NcRUtNv6Ww.png" /><figcaption>Figure: K-Means J vs k graph ( Elbow Method)</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/704/1*vMWjlSiqoHm93rEzRFSvZg.png" /><figcaption>Figure: Silhouette Coefficient</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/828/1*dtMvj7FycJt9YwOoFfvJSw.png" /><figcaption>Figure: Hierarchical clustering using ward linkage</figcaption></figure><p>Thus the inherent number of clusters in the given dataset is 2. Thus we conclude to form 2 levels — Low and High to classify our data points; based on the median value of our dependent variable i.e. ​-0.172165 (standardized value).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/849/1*t7frUGN7RaOGAuty2UAnwQ.png" /></figure><p>Based on accuracy the best performing classifiers are Linear Regression, KNN , Random Forest and LDA based classification. Linear Regression achieved an accuracy of 72.9%, KNN 80% and Random Forest 71.7%. However an important observation is that, though the LDA based classification gave good accuracy, the F1 scores are very low. This is because of low precision and sensitivity due to more cases of false positives and false negatives. Hence we select the top three classifiers Linear Regression, KNN and Random Forest for building our ensemble classifier, while discarding LDA based classifiers. In the final ensemble model we will firstly take the predictions of these three classifiers and then assign the final class based on majority voting.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/408/1*1DI-HjalpI3c2OFkTGND_Q.png" /><figcaption>Figure: Sample classification using ensemble method</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/644/1*7xAJ1--z2n9bUluIpFRJ6A.png" /><figcaption>Figure: Confusion matrix of the ensemble method</figcaption></figure><p>Accuracy : 77.6%<br>Precision: 76.1%<br>Sensitivity:78%<br>Specificity:​ 77.2%<br>F1 Score: 0.771</p><p><strong>Summary and visualization of results:</strong><br>The predictions made with the final ensemble model on the test data is represented in the following geographical heat maps.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/483/1*cIbHsYPAY0D9ZEpwP7ig2A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/607/1*GFeTbPXogYydUqOYJJS20Q.png" /><figcaption>Assam</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/663/1*zDXq9q4BdljEUtAdpQDnIg.png" /><figcaption>Uttar Pradesh</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/673/1*B97axZJQizA1farYzC9O8g.png" /><figcaption>Bihar</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/605/1*hYVWwO4Ar0dTdWvCVYgpCg.png" /><figcaption>Madhya Pradesh</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/599/1*I0oQc5MODxelI4R8huezYw.png" /><figcaption>Jharkhand</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/595/1*6OuJdUXjvFSIs8m101r4lQ.png" /><figcaption>Uttarakhand</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/615/1*7DmB_m0OEiiDC8rGCkg0Bg.png" /><figcaption>Rajasthan</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/493/1*mjVoYFAdkUxw5pHn2JKi_A.png" /><figcaption>Chhattisgarh</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/641/1*7Iaox8z4iZqhfyCErjZPOg.png" /><figcaption>Orissa</figcaption></figure><p>The heatmaps of our predictions provide an insightful visualization of the vulnerability to undernutrition among newborn children across different states. We can see that most districts of Madhya Pradesh , Jharkhand and Rajasthan fall into the ‘High’ threat level. This observation, made on the basis of our prediction, is similar to the results obtained in the National Family Health Survey (NFHS-3) 2005–6 data published by the Ministry of Health and Family Welfare, Government of India. NFHS-3 report on nutrition states that, the proportion of children under age five years who are underweight ranges from 40% in Uttar Pradesh to 60% in Madhya Pradesh. Additionally, more than half of young children are underweight in Jharkhand according<br>to NFHS-3 [​3​]. These trends have been clearly indicated by our final model’s predictions. Another observation that can be noticed is that districts of Western UP are more vulnerable compared to Eastern UP. This corresponds to the fact that there is inequality in human development parameters of western and eastern UP.[​4​]</p><h3><strong>Conclusion</strong></h3><p>Thus given any district with its above chosen socio-economic and health parameters, the final model can predict its vulnerability to undernutrition among newborn (Low or High) with 77.6% accuracy. From the data visualization using geographical heatmaps, we have also been able infer substantial insights that matches with observational data collected through other reliable surveys. Thus our final classification model is not only statistically robust but also practically deployable and capable to a large extent in explaining trends of vulnerability to undernutrition among newborn across all districts of India.</p><p>This classification of a district into different levels of vulnerability to undernutrition will provide government and non-governmental agencies the tool to mitigate the impacts of LBW. It will equip policy makers and social scientists to identify and prevent undernutrition just by observing socio-economic and health data. Thus providing a technology led approach to ensure nutrition security.</p><h3><strong>References</strong></h3><p>[1] A. Indrayan, “Epidemiological Indicators of Child Health”, ​<em>Medical Biostatistics</em>, Third Edition (2012)</p><p>[2] Nerijus Paulauskas, Ąžuolas Faustas Bagdonas, “Local outlier factor use for the network flow anomaly detection”, ​<em>Security and Communications Network</em>, (2015)</p><p>[3] Ministry of Health and Family Welfare, GoI, “Nutrition in India”, ​<em>National Family Health Survey (NFHS -3)</em> 2005–6</p><p>[4] Sumit Chaturvedi, “​Comparing Western UP and Bundelkhand : Human Development, Gender and Deprivation”, ​<em>Economic and Political weekly (EPW)</em>, Vol. 50, Issue no. 20 2015</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aa27907b6f0d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/analytics-vidhya/vulnerability-mapping-of-undernutrition-using-machine-learning-aa27907b6f0d">Vulnerability Mapping of Undernutrition using Machine Learning</a> was originally published in <a href="https://medium.com/analytics-vidhya">Analytics Vidhya</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Time series analysis of agricultural prices in India]]></title>
            <link>https://medium.com/analytics-vidhya/time-series-analysis-of-agricultural-prices-in-india-47e7ebebd8b4?source=rss-a3bb01a6f99a------2</link>
            <guid isPermaLink="false">https://medium.com/p/47e7ebebd8b4</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[timeseries]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[agriculture]]></category>
            <category><![CDATA[forecasting]]></category>
            <dc:creator><![CDATA[Abhinav Baruah]]></dc:creator>
            <pubDate>Tue, 11 Aug 2020 08:18:08 GMT</pubDate>
            <atom:updated>2020-08-23T16:46:42.860Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*m6W2WGCPnxglm3Az.jpg" /><figcaption>source: The Wire</figcaption></figure><p>Agriculture is at the center of Indian economy and any major change in the sector leads to a multiplier effect on the entire economy. With around 17% contribution to the Gross Domestic Product (GDP), it provides employment to more than 50% of the workforce. The largest industries of the country like sugar,textiles, food processing etc. are dependent on agriculture and the allied services for their raw materials. The major crops grown in India include paddy, wheat, pulses , edible oil and horticulture. Along with this poultry, fishery and animal husbandry are other major allied services. But despite the robust demand, attractive opportunities and policy support, the sector suffers from high price volatility , low income among farmers, skewed MSP rates and lack of credit facilities. Hence there is a need for a long term strategy to assess demand and supply and to forecast prices for crops.</p><p>Being a cereal-centric agriculture, the prices of paddy, wheat and pulses are of the most importance. Prices of vegetables like potato, tomato, onion also need to be closely monitored as they impact the food inflation rates in the country. Moreover the prices of 24 commodities under the MSP regime including 8 oil seeds , raw cotton, raw jute and copra are vital in deciding minimum support prices as well as deciding procurement prices for the Public Distribution System(PDS). Thus agriculture prices directly impact the income of producers and the food security of the nation.</p><p>Directorate of Economics &amp; Statistics, Agriculture (DESAg), in the Ministry of Agriculture is the prime source for agricultural statistics. Other sources include National Sample Survey Organisation(NSSO),State’s Directorate of Economics and Statistics (DES) and Central Statistical Organisation. A major part of this data is time series data, that measures agriculture prices ​over changing time frequencies — yearly, monthly, quarterly or daily. These series of prices show trends , along with seasonality and cyclic nature of agricultural prices.</p><p>These time series data associated with agricultural prices can be used for forecasting short term and long term price estimates. It can help to identify patterns and seasonal variations in prices especially with respect to production during kharif, rabi and zaid seasons. Forecasting of agricultural prices are also important for determining subsidy rates, minimum support prices, export prices etc.Thus monitoring and controlling agricultural price volatility and preplanning policies to mitigate any adverse situations. Time series of crop prices can also help to understand, interpret and explain instances of economic crisis in the agricultural sector. Intervention analysis of such time series can also provide us insights into understanding the impact of sudden price changes over other sectors and overall economy. Moreover agriculture price forecasts are essential in determining risks in agri-commodity futures and options trading markets.</p><p>An important component of time series data is trend and agricultural price data often show deterministic trends. Trend analysis can be performed by using ​Moving Average Method ​and ​Exponential Smoothing Method​. In the Moving Average Method, the first average is the mean of the prices in the first ​m period. The second average is the mean of the prices of ​m data points starting from the second price data up to (​m + 1) th period. Similarly, the third average is the mean of the prices of ​m data points from the third to (​m + 2) th period and so on. This new set of data is then used to determine the trend.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/683/1*qUO8ki6bz2WBXImxZ2SF6A.png" /></figure><p>The Exponential Smoothing Method gives larger weights to recent price observations compared to past observations. Such that the weights decrease exponentially.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/667/1*6or2IYcHo2GS7tPuYhmkPg.png" /></figure><p>Another component of time series, that is often present in agricultural price data is seasonality. This is evident from the fact that the harvesting of crops takes place over a certain period of time in a year and the prices fluctuate accordingly. It is thus necessary to identify the seasonality and obtain the<br>deseasonalized series for further model development. The methods used are the ​Dummy Variable method ​and ​Ratio to Moving Average​ methods.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/612/1*70qJgdwFKl7PUNQZTN3PzQ.png" /></figure><p>Before forecasting the price of agriculture produce, it is essential to identify the presence of non-stationarity. This is necessary to avoid spurious and misleading model estimates. The most common test used for the purpose is ​Augmented Dickey Fuller(ADF) unit root test. The framework of ADF methods is based on analysis of the following model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/347/1*9QENJpYPq5EZYq4DALR9Lg.png" /></figure><p>If the time series data is non-stationary then we make it stationary by differencing and again performing the ADF test. This step is repeated until we get a stationary time series. After getting a deseasonalized stationary series without having a deterministic trend, we can proceed to identification of the model. The model can be either ​AR(p) / MA(q) / ARMA(p,q)​. To determine the model and its order we calculate the sample ACF ,sample PACF and use the ​L Jung Box Test to conclude the final model. The parameter estimation can be done based on the MLE/ Yule Walker equation/OLS approach. ARIMA models are fitted and accuracy of the model is tested on the basis of residual diagnostics. The L Jung Box test is applied on the residuals to determine whether they are white noise. The final time series model is thus ready for forecasting and prediction.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/373/1*D6Oym20ggVKPyVZGvV2YWw.png" /></figure><p><strong>ILLUSTRATIONS</strong><br>1. Price forecast of paddy based on ARMA(1,1) model</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/721/1*OczbLSPsJOomgTEHgGxomQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1017/1*Au9TfM-d7ylS8ueS4GytiQ.png" /></figure><p>Based on the ACF and PACF correlogram, a time series ARIMA(1,1,1) model has been built to forecast paddy prices for the future.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/534/1*lkc_gVQcamukKnq0fecQiA.png" /></figure><p>2. Price forecast of maize based on MA(1) model</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/513/1*f5mslB5zaR8Q2kanm8ED2w.png" /></figure><p>3. Price forecast of pigeon pea based on AR(1) model</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/553/1*Vq6NHyC02SoNYv9fS6WVLQ.png" /></figure><p><strong>CONCLUSION</strong></p><p>The time series AR/MA/ARMA models serve as a good methodology for predicting the magnitude of any price variable. Its strength lies in the fact that the method is effective in any time series on agricultural prices with a pattern of change. Fluctuation in prices have increased significantly with a lot of uncertainty in profitability of growing agricultural crops. Under this scenario, forecasting future prices and disseminating the same to farmers to facilitate informed decisions by farmers before sowing is important. Forecasting of prices can help the farmers to decide the area allocation for a particular crop and marketing of agricultural produce. The forecasted values are almost similar to actual prices but like any other method, this technique also does not guarantee perfect forecasts. Nevertheless time series analysis is handy and can be successfully implemented using R and Python programming to get desired results.</p><p><strong>REFERENCES</strong><br>1. V. Jadhav , B. V. Chinnappa Reddy , and G. M. Gaddi(2017), Application of ARIMA Model for Forecasting Agricultural Prices ,​J. Agr. Sci. Tech. Vol. 19: 981–992<br>2. Ashwini Darekar · A. Amarender Reddy(2017),Price forecasting of maize in major states, ​Maize Journal 6(1&amp;2): 1–5<br>3. Ashwini Darekar · A. Amarender Reddy(2017), Price forecasting of pulses: case of pigeonpea,​Journal of Food Legumes 30(3): 42–46</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=47e7ebebd8b4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/analytics-vidhya/time-series-analysis-of-agricultural-prices-in-india-47e7ebebd8b4">Time series analysis of agricultural prices in India</a> was originally published in <a href="https://medium.com/analytics-vidhya">Analytics Vidhya</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>