Бидний тухай
Багш ажилтан
Цахим сургалт нь зөвхөн Ковидын цар тахлын нөлөөг үл тооцсон ч сүүлийн жилүүдэд улам түгээмэл болж буй талаар эрдэмтэд судалгааны ажилдаа тэмдэглэсээр байна. Орчин үеийн энэхүү хандлагын дагуу магадлал, статистикийн хичээлийн сургалтын цахим хэрэглэгдэхүүн агуулсаны зэрэгцээ цахим шалгалтын модулыг багтаасан цогц систем хөгжүүлж буй нь www.magadlal.com веб аппликейшн юм. Цахим сургалтад тулгардаг бэрхшээлүүдийн нэг бол суралцагчийн мэдлэг чадварыг үнэн бодитой шалгах нэн ялангуяа шалгалтын үеэр бусдаас хуулах явдал байдаг ажээ. Иймд тус веб аппликейшны шалгалтын модулын тусламжтай авсан цахим шалгалтын үр дүн дээр хуулалтын ул мөрийг мөшгөх судалгаа явуулж улмаар үүнд үндэслэн хуулалтын эсрэг арга техникүүдийг тус модулд тусгах зэрэг ажлыг үе шаттай хийн зарим ахиц дэвшилд хүрээд байна. Ийнхүү уг илтгэлээр цахим шалгалтын хуулалтыг мөшгөсөн хийгээд цахим шалгалтын хуулалтын эсрэг зарим арга хэмжээний үр нөлөөг судлахад гарсан үр дүнгээс танилцуулна.
Water quality monitoring is one of the main problems for Mongolia since people and livestock directly use surface and shallow well water for drinking. Water quality could be evaluated by bacterial abundance and diversity changes. There are lots of studies on microbial abundance and diversity along rivers. However, most of them are conducted in developed countries. Therefore, it is needed to investigate the bacterial abundance and diversity changes in our rivers. We will introduce how did we detect the bacterial abundance and diversity changes along the Kharaa River. In terms of diversity analysis, we did not find significant changes, especially at the broader taxonomic categories such as phylum, subphylum, class, order, and family. But we found significant shifts in the OTU (Operational Taxonomic Unit) abundance, especially by detecting the most trendy (with increasing and decreasing abundance along the river) OTUs along the Kharaa River. In the next step, we will analyze what environmental factors are shaping these shifts.
R is the most powerful programming language and the most popular environment for statistical computing and visualization. The National Statistical Office (NSO) of Mongolia is not an agency of the government; it is an independent agency under the supervision of the Parliament. NSO provides official statistical reports and data sets for the government and international organizations. Also, NSO distributes several data sets freely available for open access. Open data is a revolutionary trend in Mongolia nowadays. NSO is conclusively leading this trend in the country. NSO1212 is an R package, contains functions for effortlessly accessing such data through NSO’s API. Importing data with the package has an advantage in filtering and tidying data for further processing and analysis. In this talk, we introduce our developed package NSO1212 and consider several issues on the NSO’s API server that can cause significant inconvenience to ordinary users.
In this research work, we propose a novel discrete probability distribution which is an extension of the discrete triangular distribution. The proposed distribution can be used for a wide range of random variables which are triangular, upside-down triangular, and asymmetric standard triangular distributed due to its flexible tails. Thus it is called flexible tailed discrete triangular distribution. Also, this distribution is not only related to degenerate, uniform, and Bernoulli distribution but also the extension of the other existing discrete triangular distribution and its various forms. The most important advantages of such distribution are covering various discrete distributions having finite support and easy to fit on real data.
In this talk we introduce a novel similarity measure for categorical random variables and a statistical test for identification of Mongolian hybrid racing horses from native. There are negative facts about hybrid horses hereto it’s vulnerable against diseases and losses its native behavior. However, there is no research work about hybrid horses except Bilegdemberel Banzragchs thesis. In such research work, only phenotype of racing horses is considered. Federation of Mongolian Horse Racing Sport and Trainers is trying to classify hybrid and native racing horses which compete at the national level race such as Naadam. The classification rule is based on crest of a horse. The racing horse trainers complain about this simple rule. In population genetics, microsatellite markers are used to identify individuals and measure difference or relation between populations and individuals. Thus, we developed a test which is based on 19 microsatellite markers.
In this research work, we propose a novel discrete probability distribution which is an extension of the discrete triangular distribution. The proposed distribution can be used for wide range of random variables which are triangular, upside-down triangular and asymmetric standard triangular distributed due to its flexible tails. Thus it is called flexible tailed discrete triangular distribution. Also this distribution is not only related to degenerate, uniform and Bernoulli distribution but also the extension of the other existing discrete triangular distribution and its various forms. The most important advantages of such distribution are covering various discrete distributions having a finite support and easy to fit on real data.
Generally, currency exchange rate forecasting models are essential for normal economic conditions. However, Mongolia's economy is based on mining and livestock agriculture. Also the other main factor of the economy is foreign direct investment. Thus, Mongolia's economy is vulnerable and exchange rate of the national currency Tugrug (MNT) is unstable. Accordingly, currency exchange rate needs to be observed everyday. We developed the currency exchange rate alert system for MNT and USD with machine learning approaches by using a linear regression model. The system detects a structural or an accidental change of exchange rate by using a statistical test or a confidence interval, then sends a notification to email addresses of subscribed users. Currently, our developed system is working on the cloud server and its home page address is www.magadlal.com/xrate.
Extracting a group of genes whose activation simultaneously produces significant effects on cancer progression is of great importance in understanding the details of control mechanisms in cancer-responsive biological processes. Network structures have been often used to represent these molecular activities based on an accurate estimation of behavioral relationships that is a whole set of interacting neighbors and local patterns, responsive under investigated conditions. Here, we propose a novel method, so-called Weighted Maximum Clique Tree (WMCT), to identify a particular subset of genes forming a cancer-specific sub-network. We first estimate a gene-network using a Kullback-Leibler divergence with a sparse covariance matrix by measuring the differential co-expression signatures across physiological conditions. Then an existing protein-protein interaction network information is combined to the activation of gene-network for constructing a background cancer-responsive network. Inspired by an integer linear programming formulation, the densest part of the network is obtained to represent the cancer-specific sub-network. We applied the WMCT method to both simulated data and a real data set of prostate cancer. WMCT successfully identified a large fraction of well-known oncogenes in prostate cancer and the subset of genes were enriched in cancer-related pathways and biological processes with significant p-values. Compared with several existing methods, WMCT returned better performances in simulated data sets. These results demonstrate that the proposed method can efficiently identify a particular subset of genes relevant under investigated condition.
Сүүлийн жилүүдэд ШШҮХ-нд ялтны ДНХ-ийн 7000 гаруй биологийн дээж биетээр ба цаасан баримтаар хуримтлагдсан. Энэхүү судалгааны хүрээнд бид MonDIS нэртэй үндэсний хэмжээний анхны ДНХ-ийн мэдээллийн сан, хайлтын системийг үүсгэн байгуулсан ажлын үр дүнг тайлагнана. Мөн энэ чиглэлийн судалгаатай холбоотой хуримтлуулсан туршлага, цаашид гүйцэтгэх ажлыг танилцуулна.
Genomic activations in cancer are a mixture of driving events that promote cancer progression and passenger events that represent a large fraction of random somatic alterations. Extract- ing a group of genes whose activations simultaneously produce significant effects on cancer development is of great importance in understanding the details of control mechanisms in cancer-responsive biological processes. Here, we propose a novel method, so-called Weighted Maximum Clique Tree (WMCT), to identify a condition specific sub-network. We first con- struct a gene-network using a Kullback-Leibler divergence with a sparse covariance matrix by measuring the differential co-expression signatures across physiological conditions. Inspired by an integer linear programming formulation, the densest part of the network is obtained to rep- resent the condition specific sub-network. We applied the WMCT method to both simulated data and a real data set of prostate cancer.
Here, we propose a novel method, so-called Weighted Maximum Clique Tree (WMCT), to identify a cancer-responsive sub-network. We first construct a gene-network using a modified Kullback-Leibler divergence with a filtration for expectations and a sparse covariance matrix by measuring the differential co-expression signatures across physiological conditions. Inspired by an integer linear programming formulation, the densest part of the network is obtained to represent the cancer-responsive sub-network. We applied the WMCT method to both simulated data and a real data set of prostate cancer.
Monte Carlo simulation is the most popular method for the problems which are difficult to solve by using the other approaches. It requires random values generated from unknown distributions. It is not practically easy to find an analytic form of the distribution or to fit a probability distribution. In such case, empirical distributions are widely used. However it has a limitation of handling a sampling error due to non-smoothness shape of distribution. To overcome the issue, a kernel density estimation procedure can be used to smooth empirical density functions. The second challenge is to generate random values from a given distribution. Although var- ious methods for simulating random variables have been developed, a generalization is lacking due to high dimensionality and distribution types. The Accept-Reject method can be used to simulate any distributions. But its efficiency depends on a choice of the instrumental or candidate random variable. One possible approach to increase the efficiency is to use the best fitting distribution as the instrumental distribution. In this presentation, we introduce an R package providing functions for finding the best fitting instrumental distribution and, generating random values from a specific distribution which is determined from data by using the kernel density estimation and the Accept-Reject method. The package was submitted to the CRAN and available on the page https://cran. r-project.org/package=simukde.
Monte Carlo simulation is the most popular method for the problems which are difficult to solve by using the other approaches. It requires random values generated from unknown distributions. It is not practically easy to find an analytic form of the distribution or to fit a probability distribution. In such case, empirical distributions are widely used. However it has a limitation of handling a sampling error due to non-smoothness shape of distribution. To overcome the issue, a kernel density estimation procedure can be used to smooth empirical density functions. The second challenge is to generate random values from a given distribution. Although var- ious methods for simulating random variables have been developed, a generalization is lacking due to high dimensionality and distribution types. The Accept-Reject method can be used to simulate any distributions. But its efficiency depends on a choice of the instrumental or candidate random variable. One possible approach to increase the efficiency is to use the best fitting distribution as the instrumental distribution. In this presentation, we introduce an R package providing functions for finding the best fitting instrumental distribution and, generating random values from a specific distribution which is determined from data by using the kernel density estimation and the Accept-Reject method. The package was submitted to the CRAN and available on the page https://cran. r-project.org/package=simukde.
