تحلیل مصرف آب شهری (خانگی) شهرستان بابل با استفاده از روش های داده کاوی

مسئله کمبود آب در ایران با توجه به قرارگیری آن در منطقه خشک و کم آب خاورمیانه و روند نسبتاً سریع افزایش جمعیت شهری و مصرف بی‌رویه آب، هر روز ابعاد جدی‌‌تری به خود می‌‌گیرد. این مسئله و همچنین فقدان الگوی مشخص از نحوه و میزان مصرف آب در کشور، استفاده از سیستم‌‌های پردازش اطلاعات را در مدیریت منابع آب ضروری می‌‌نماید. کاوش قوانین وابستگی می‌‌تواند به عنوان یکی از روش‌‌های مطلوب جهت بهینه‌‌سازی و تحلیل عوامل موثر در میزان مصرف آب مورد استفاده قرار گیرد. در این مطالعه از کاوش قوانین وابستگی و الگوریتم درخت تصمیم به منظور تحلیل داده‌‌های موجود در زمینه مصرف آب شهری محله‌‌های شهرستان بابل استفاده شده است. پارامترهای مکانی و اجتماعی اقتصادی  مانند فاصله از رودخانه بابل‌رود، فاصله از راه‌‌های اصلی، ویلایی یا آپارتمانی بودن، مساحت فضای سبز خانگی،  تعداد واحدهای مسکونی، تراکم جمعیت، درصد پیر یا جوان بودن، متوسط تعداد افراد هر خانواده و مساحت حیاط ساختمان انتخاب شدند. با استفاده از کاوش قوانین وابستگی به کشف ارتباط میان میزان مصرف آب و پارامترهای موردنظر پرداخته شده است. با بهره‌‌گیری از یک درخت تصمیم، عوامل مرتبط با مصرف، طبقه‌‌بندی و میزان مصرف آب بخشی از مشترکین به عنوان داده‌‌های تست پیش‌‌بینی شده است. به کمک تعیین نواحی با بیشترین میزان مصرف آب و الگوی مکانی توزیع مصرف آب خانوار، این قوانین ارزیابی شده‌‌اند. در این پژوهش پارامترهایی که رابطه عکس یا مستقیم با میزان مصرف آب خانوار دارند و همچنین پارامترهایی که اهمیت بیشتری نسبت به سایر عوامل دارند، مشخص شده‌‌اند. با بررسی میزان مصرف آب خانوار، محله‌‌های با بیشترین میزان مصرف که محله‌‌های ساحلی رودخانه بابلرود هستند، به عنوان محله‌‌های پرمصرف شناسایی شده‌‌اند. ساختمان‌‌ها در این محله‌‌ها، ویلایی و دارای تراکم جمعیت پایین‌‌تر و مساحت فضای سبز و حیاط بالاتری نسبت به سایر محله‌‌ها هستند.

Analysis of urban water consumption in Babol County using data mining methods

Extended Abstract Introduction and Objective Due to the location of Iran in dry regions of the Middle East, and also because of the rapid increase in its urban population and water consumption, every day the issue of water scarcity becomes more severe in Iran. In recent years, Iran has faced serious water scarcity and excessive consumption of water resources. Therefore, patterns of urban water consumption, different geographic, spatial, demographic, social, and economic parameters, and the relation between these parameter and water consumption are considered to be among important issues affecting management of water resources. The present study seeks to investigate and analyze the spatial pattern of domestic water consumption in Babol County, and also to identify parameters affecting the pattern of water use. This is achieved by extracting association rules from some spatial and socioeconomic parameters and based on the water use level in this County. The study also aims to determine regions with high/low level of water use, investigate spatial distribution of water consumption and finally, identify and categorize parameters affecting domestic water consumption at neighborhood level in this County using Decision Tree model.   Materials and methods Data: Domestic water consumption data, census data, spatial and socioeconomic parameters such as distance from main roads, distance from Babolrood, total area of garden and green space in each building, building site and standing property (total area of house yard), population density, total number of houses vs. apartments, number of housing units, average number of people per household, percentage of young/old people per household were extracted from the Statistical Center of Iran for the time period of 2011 to 2016. Then, these data were used to analyze urban water consumption in Babol County. Methods: Apriori algorithm a data mining algorithm used to extract association rules has been used to discover and extract relationships between different spatial socioeconomic parameters and domestic water consumption patterns. Moreover, a decision tree has been developed which takes advantage of these parameters to predict domestic water use.   Results and Discussion  Results indicated that number of houses, number of household members, green space in each house, total area of house yard and distance from main roads are directly related with the household water consumption. On the other hand, population density, percentage of youth population, number of residential units and distance from Babolrood River are inversely related to domestic water consumption. Among all parameters considered in the present study, total area of house yard, distance from Babolrood River, number of residential units and number of household members exhibited a stronger relationship with water consumption. Thus, they were located on higher branches of the final decision tree. Additionally, results of global Moran’s I index indicated that there exists a spatial autocorrelation among household water consumption data. Moreover, this index indicated the clustered nature of residential water consumption distribution in Babol County. Also, spatial distribution of domestic water consumption in this County demonstrated that western and coastal areas with minimum distance from Babolrood River have the highest level of domestic water consumption. Therefore, it can be concluded that with an increase in distance from Babolrood River, domestic water consumption decreases. Only terraced and semidetached houses exist in these neighborhoods. Thus compared to other neighborhoods, they have a lower population density, larger green space and larger yard.   Conclusion and Future Works   The present study applies Apriori algorithm to extract association rules and discover the relationship between spatial and socioeconomic parameters and domestic water consumption. Results indicated that spatial and socioeconomic parameters affect the spatial distribution of domestic water consumption in Babol County. Developing a decision tree, parameters associated with domestic water consumption were categorized and amount of water consumption was predicted. Extracted rules predicted domestic water consumption of test data with an accuracy of 75%. In this study, global Moran’s I index indicated the existence of a spatial autocorrelation among water consumption data. It also proves the clustered nature of domestic water consumption distribution in the study area. Additionally, spatial distribution of domestic water consumption in Babol County indicated that western and coastal neighborhoods have the highest level of domestic water consumption, while southern neighborhoods of Babol County have the lowest level of domestic water consumption. Model developed in the present study provides an opportunity for analyzing and predicting the level of water consumption. This will make planning for the reduction of water consumption and management of water resources possible. We suggest that future works evaluate the effect of other spatial and socioeconomic parameters such as water cost and educational status of household members in a longer period (more than 5 years) to improve the accuracy of the model.