Exposure to the natural radioactive background radiations is unavoidable on the planet earth, which results in adverse health effects, even leading to the fatal cancer. In this context, an investigation of natural background radiation was carried out for environmental and building material samples in Muzaffarabad and Poonch divisions of the State of Azad Jammu and Kashmir. Multivariate techniques were employed for radiation data analysis and assessment of the environmental radiological status of the study area along with scrutinizing the suitability of the building materials used for construction purposes. In initial phase of the study, spring and Well water samples were collected from Muzaffarabad city and its outskirts, whereas sand and sediment samples were collected from riverbanks of Jehlum, Neelum and Kunhar rivers. Soil and rock samples were taken from Leepa Valley. Decorative stone samples were taken from different geological formations of Azad Jammu and Kashmir, whereas gamma dose rates in air were measured in Poonch division of the state and finally machine learning technique was used to determine the best suitable radionuclide for classification of rocks. Spring and well water samples from the study area were investigated for water borne radon activity (WBRnA). WBRnA was measured in a total of 101 water samples including 60 springs and 41 Well water samples, using RAD7 alpha spectrometer. Mean values of WBRnA in spring and Well water samples were found to be 10.16 ± 2.42 and 4.21 ± 0.13 Bq L-1, respectively. It is found that 33.33% of spring and 7.32% of Well water samples values were above the recommendation level (̴ 11.1 Bq L-1) proposed by United States Environmental Protection Agency (USEPA). Mean values of inhalation and ingestion doses were computed and found to be 0.26 ± 0.0039 mSv y-1 and 2.13 ± 0.033 mSv y-1, for spring water and 0.106 ± 0.00324 mSv y-1 and 0.884 ± 0.027 mSv y-1 for Well water samples, respectively. Total annual effective dose due to WBRn was found to be 2.16 ± 0.008 mSv y-1 and 0.89 ± 0.027 mSv y-1, for spring and Well water samples respectively. Mean value of excess lifetime cancer risk (ELCR) for spring and Well water samples, was computed to be 7.56 × 10-3 and 3.13 × 10-3, respectively, which are much higher than the upper bound of 0.1 × 10-3 for drinking water, as proposed by USEPA. Radon exhalation rate (RER) was measured for sand and sediment sample and was found to be 335 mBq m-2 h-1 and 259.21 mBq m-2 h-1, respectively. Gamma dose rates in air were measured, in Poonch Division, using Thermo-luminescent dosimeters (CaF2: Dy (TLD-200) card dosimeters) and NaI(Tl) survey meter for the estimation of excess lifetime cancer risk (ELCR) along with inter-comparison of active and passive techniques. An empirical relationship was developed to get annual gamma dose rate estimates from TLDs based six-month data. Mean annual effective dose equivalent (AEDE) was obtained as 0.298 mSvy-1 from TLDs and 0.175 mSv y-1 from NaI(Tl) detector; both values are less than the worldwide average annual effective dose (0.48 mSv y-1). Gamma doses obtained by TLDs are higher than those measured by the survey meter. Mean values of ELCR obtained from TLDs and the survey meter results, were found to be 9.85 ×10-4 and 5.77 ×10-4, respectively and these values are higher than the worldwide average value of 0.29 ×10-3. High Purity Germanium (HPGe) based gamma spectrometric system was used for the measurement of radioactivity in different environmental samples. Mean gamma activities of 226Ra, 232Th, 40K in sand; sediment; soil; rock; and decorative stone samples were found to be 48.25 ± 1.77, 44.58 ± 3.34 and 239.92 ± 22.73 Bq kg-1; 36.91 ± 1.12, 61.57 ± 2.81and 557.33 ± 19.81 Bq kg-1; 31.25±0.46, 44.1±1.07 and 575 ± 8.89 Bq kg-1; 28.46 ± 0.45, 48.63 ± 1.12 and 666.7 ± 9.39 Bq kg-1; and 37.32 ± 0.38, 38.57 ± 0.74 and 465.62 ± 5.35 Bq kg-1, respectively. Radium equivalent activity (Raeq) for these types of samples was found to be 130.47 ± 8.29; 167.88 ± 6.66; 143.16; 142.93; and 126.30 ± 1.79Bq kg-1, respectively. In all cases, Raeq is less than allowed upper limit of 370 Bq kg-1 as given in Organization of Economic Cooperation and Development (OECD) report (1979). Mean estimates ofhazard indices, which include external hazard index (Hex), internal hazard index (Hin), gamma index (Iᵞ) and alpha index (I a ) in all cases were found to be less than unity except representative index (Ir) having mean value (1.23 ± 0.04) greater than unity for sediment samples only. Gamma ray absorbed dose rate (mean values) in sand; sediment; soil; rock; and decorative stone samples was estimated to be 59.22 ± 3.78; 77.49 ± 3.04; 69.78; 70.01; and 59.09 nGy h-1, respectively. For the sand samples, dose rates are comparable with the worldwide average value of 59 nGy h-1, whereas for all other types of samples, dose rates are higher than this value. Mean values of annual effective dose equivalent (AEDE) for these material samples were computed to be 0.363 ± 0.023; 0.48 ± 0.02; 0.410 ± 0.008; 0.436 ± 0.008; and 0.36 mSv y-1, respectively. These AEDE values are comparable with worldwide average value of 0.48 mSv y-1 but lower than the ICRP (ICRP103, 2007) proposed Reference Level (1-20 mSv y-1). Annual gonadal equivalent dose (AGED) was also calculated for the same sample types and found to be 410.77 ± 26.56; 546.445; 461.351 ± 8.683; 500.60 ± 9.03; and 431.11 ± 5.92 µSv y-1, respectively. AGED in all cases are higher than the UNSCEAR 2000 proposed limit of 300 µSv y-1. Mean values of ELCR for sand; sediment; soil; rock; and decorative stone samples were assessed to be (1.199 ± 0.077) ×10-3; 1.57 ×10-3; (1.354 ± 0.025) ×10-3; (1.438 ± 0.027) ×10-3; and (1.237 ± 0.017) ×10-3 respectively. Although in all cases, ELCR values are higher than the worldwide average value of 0.29×10-3. Finally, we have used supervised learning algorithms for building prediction and classification models. Radionuclide data, of each rock samples, has been randomly divided into train and test set of different weightages. Our aim of this part of study was to asses which one of radionuclide will be best suitable for the purpose of classification of rocks. This study has identified the best suitable ML method for classification of rock types, using radionuclide data set, in terms of accuracy and other performance measures. In public health perspective, it is concluded that health risks to the population of the study area due to NORMs in the investigated environmental sample are generally insignificant. However, the inhabitants of the study area, are at risk to a certain degree of developing cancer over a period of their life time. In addition, some sediment samples have revealed elevated levels of hazard indices and effective doses, therefore, such sediments should be prohibited to be used as buildingnmaterials.