Water quality mapping in Pontianak City using multiple discriminant analysis

• Main Authors: Kusnandar, Dadan, Debataraja, Naomi Nessyana, Rizki, Setyo Wira, Saputri, Ernita
• Format: Conference Proceeding
• Language: English
• Subjects: Biochemical oxygen demand; Classification; Discriminant analysis; Geographic information systems; Independent variables; Mapping; Multivariate analysis; Normal distribution; Oxygen; Quality standards; Residential areas; Water hardness; Water pollution; Water quality
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0016809

Discriminant analysis is a common tool for classifying objects into two or more classes. It is generally assumed that the data follow a multivariate normal distribution. This research aimed to determine a model to classify the criteria for water pollution in the residential area of Pontianak City. Multiple discriminant analysis was used for these purposes. The data were collected from 42 samples that can be classified into three criteria, i.e., meeting the quality standard, lightly polluted, and moderately polluted. Independent variables are chemical content in the sample, i.e., fluoride ( x1 ), water hardness level ( x2 ), nitrite ( x3 ), dissolved oxygen ( x4 ), and biochemical oxygen demand (x5 ). Of the 42 samples, eight samples were classified as meeting the quality standard, 24 samples as lightly polluted, and ten samples as moderately polluted. Only two variables were significant in the model, i.e., water hardness and biological oxygen demand. The discriminant model obtained was d̂12 = −4,371 + 0,066x2 + 0,098x5 , and d̂13 = 0,462 −0,050x2 + 0,080x5. The resulting Apparent Error Rate of the model was 38.1%. The sample points were also visualized into a map with the Geographically Information System (GIS).