Agglomerative Clustering Method Research Articles

Effects of the Disbursement of EU Cohesion Policy 2014–2020 Funds on Improving the Energy Efficiency of Buildings in Poland and Germany

Improving the energy efficiency of buildings (IEEBs) is crucial to achieving the European Union’s ambitious climate and energy goals. To channel these efforts, new policies, legislation, sources of funding, mechanisms for distributing funds, programs, and support tools are being introduced. The purpose of this study is to demonstrate the effects of the disbursement of EU Cohesion Policy 2014–2020 funds on IEEBs in Poland and Germany, filling the research gap regarding the regional differentiation of investments, beneficiaries, and the spatial distribution of subsidies and projects. An analysis was made of 7818 projects on IEEBs implemented under regional operational programs from European Regional Development Funds under Thematic Objective 4. This research was conducted for two main variables: (1) beneficiaries (8 categories) and (2) investments (8 categories). Based on this division, using a cluster analysis through the use of Ward’s agglomerative hierarchical clustering method, groupings were made in relation to beneficiaries (6 groups) and investments (6 group). The results allowed us to identify regional differences and similarities in the approach to the disbursement of funds on IEEBs, to indicate the main beneficiaries, to highlight trends in this regard, to demonstrate the impact of changes in territorial affiliation on investment targeting, and, furthermore, to capture the variation in the sophistication of the implementation of EU IEEB-related policies and legislation. The results provide a well-documented voice in the discussion of the effects of distributing EU funds for IEEBs in the regions of Poland and Germany and provide a basis for further comparative research.

Mapping phenoregions and phytoplankton seasonality in Northeast Pacific marine coastal ecosystems via a satellite-based approach

Phytoplankton phenology describes yearly algal growth cycles and characterises the timing, duration, and magnitude of bloom occurrences. This study used satellite chlorophyll-a data from 1998 to 2020 and the Hierarchical Agglomerative Clustering method to define phenoregions based on phytoplankton phenology spatial patterns over the British Columbia and Southeast Alaska coastal oceans. The defined phenoregions were used to simplify the spatial complexity of the heterogenous study region and thus better describe phytoplankton seasonality across the target area. The cluster analysis allowed the delineation of four coherent regions: two coastal regions and northern and southern shelf/offshore regions. Results showed that each phenoregion had distinguishable phytoplankton phenological characteristics, likely due to different physical forcings acting in these areas. Moreover, the interannual variability of the spring bloom initiation was evaluated considering interactions between sea surface temperature (SST) anomalies and the El Niño Southern Oscillation Index (ENSO). Early spring blooms were associated with positive SST anomalies and El Niño conditions; conversely, average or late spring blooms occurred in years with negative SST anomalies and La Niña conditions, with the strongest relationship occurring in the southern shelf/offshore phenoregion. This study provided new insights into the regionalisation of the British Columbia and Southeast Alaska coastal oceans based on phytoplankton phenology patterns. Given the critical role of phytoplankton as the base of the marine food web, such phenoregions have implications for regional zooplankton biomass and fish production. The link between phytoplankton phenology and climate drivers points to the importance of environmental change in phytoplankton bloom dynamics. Further research into the connection between phytoplankton bloom indices and zooplankton community structure and production would be an important step towards using these indices for ecosystem monitoring and fisheries management.

The approach development of data extraction from lambda terms

The study's object is the process of extracting the characteristics of lambda terms, which indicate the optimality of the reduction strategy and increase the productivity of compilers and interpreters. The solution to the problem of extracting specific strategy priority data from lambda terms using Machine Learning methods was considered. Such data was extracted using the large language model Microsoft CodeBERT, which was trained to solve the problem of summarizing the software code. The resulting matrices of embeddings were used to obtain vectors of average embeddings of size 768 and a latent space of size 8 thousand. Further, vectors of average embeddings were used for cluster analysis using the DBSCAN and Hierarchical Agglomerative clustering methods. The most informative variables affecting clustering were determined. Next, the clustering results were compared with the priorities of reduction strategies, which showed the impossibility of separating terms with RI priority. A feature of the obtained results is using machine learning methods to obtain knowledge. The clustering results showed many of the same informative variables, which is explained by the similar shape of the obtained clusters. The results of comparing the clustering values with the real priority are explained by the impossibility of clearly determining the priority and the use of the Microsoft CodeBERT model, which was not trained for the analysis of lambda terms. The proposed approach can find application in the development of compilers and interpreters of functional programming languages, allowing to analyze the code and extract important data to optimize the execution of programs. The obtained data can be used to develop rules aimed at improving the efficiency of compilation and interpretation

Design and development of an efficient real-time moving object tracking system in a wireless-controlled automated vehicle

This study aimed to apply a microprocessor on a wireless automated vehicle to achieve real-time tracking of moving objects. The targets captured by the camera on the vehicle were first separated from their background through background subtraction. Next, morphological processing was performed to remove unnecessary information. An enhanced seeded region growing method was used to achieve image segmentation by labeling and segmenting the targets effectively, thus enhancing the accuracy and resolving the problem of object concealment. The corresponding red, green, and blue colors of each target were calculated through a color space, which was then converted into an enhanced luminance/chroma blue/chroma red (YUV) color space for color histogram modeling and storage, so as to increase the system’s tracking speed. The enhanced YUV colors also achieved accurate tracking in dark places. After inputting the next image, an enhanced agglomerative hierarchical clustering method was used to agglomerate and connect pixels with the same YUV for tracking. A proportional-integral-derivative controller controlled the motors on the camera lens and the vehicle so that the target could be tracked properly in real-time. The experimental results revealed that our proposed tracking method performed better than conventional tracking methods.

Identification of Digital Divide across Indonesian Provinces: the Analysis of Key Factors

Despite the economic and societal benefits of digitalisation and digital transformation, it is necessary to map country's digital conditions and identify the digital divide to formulate an effective strategy. The digital divide should be measured periodically to monitor progress and determine continuous improvement. This paper identifies the current digital divide among provinces in Indonesia. The study uses the hierarchical agglomerative clustering method based on The Indonesian Digital Society Index data from the Ministry of Communication and Informatics. It also analyses some key factors of the digital divide based on data from the Indonesian Bureau of Statistics using the multiple linear regression model. The results show three types of the digital divide across Indonesian provinces related to access, usage, and outcomes of information and communication technology. Gross Regional Domestic Product per capita, Wage/Salary, Proportion of Formal Labor, and Size of the Working-Age Population are identified as factors significantly affecting the digital divide.

Klasterisasi Kabupaten/Kota di Provinsi Papua Berdasarkan Indikator Pembangunan Manusia Tahun 2022

The Human Development Index (HDI) is a human development metric empowered by several main elements that reflect the quality of life, namely HLS, RLS, PKK, and UHH. High HDI is one of the basic capital for a country's sustainable development. The improvement of human development in Indonesia continues to increase every year. However, there is still uneven development between regions in Indonesia. The HDI of Papua Province has always been the lowest in Indonesia in the last 10 years and each indicator is still unevenly distributed between districts/cities. This needs to be a special concern by the government. Thus, this study was carried out with the aim of grouping districts / cities in Papua Province based on human development indicators with the application of the agglomerative hierarchy cluster method. Based on the highest correlation value, Average Linkage is the best method with the optimal number of clusters formed is five clusters (Very High, High, Medium, Low, and Very Low) according to the Calinski-Harabasz Index test conducted. The findings of this study can be the basis for designing equitable development policies in districts/cities in Papua Province.

Assessment of American Bullfrog (Lithobates catesbeianus) spreading in the Republic of Korea using rule learning of elementary cellular automata

The spread of American Bullfrog has a significant impact on the surrounding ecosystem. It is important to study the mechanisms of their spreading so that proper mitigation can be applied when needed. This study analyzes data from national surveys on bullfrog distribution. We divided the data into 25 regional clusters. To assess the spread within each cluster, we constructed temporal sequences of spatial distribution using the agglomerative clustering method. We employed Elementary Cellular Automata (ECA) to identify rules governing the changes in spatial patterns. Each cell in the ECA grid represents either the presence or absence of bullfrogs based on observations. For each cluster, we counted the number of presence location in the sequence to quantify spreading intensity. We used a Convolutional Neural Network (CNN) to learn the ECA rules and predict future spreading intensity by estimating the expected number of presence locations over 400 simulated generations. We incorporated environmental factors by obtaining habitat suitability maps using Maxent. We multiplied spreading intensity by habitat suitability to create an overall assessment of bullfrog invasion risk. We estimated the relative spreading assessment and classified it into four categories: rapidly spreading, slowly spreading, stable populations, and declining populations.

Comparative Evaluation of Six Agglomerative Hierarchical Clustering Methods With a Robust Example

The agglomerative hierarchical clustering methods are the most popular type of hierarchical clustering used to group objects in clusters based on their similarity. The methods are represented by a bottom-up approach where each object starts in its cluster and pairs of clusters are merged as it moves up the hierarchy. In this paper, we present six agglomerative hierarchical clustering methods namely: the single linkage method, complete linkage method, average linkage method, centroid method, median method, and Ward’s method. We also evaluated how these methods work on a practical basis using a matrix of distance pairs of five points. It was observed that the single linkage method through its dendrogram produced the most similarity measure between x_i and x_j, while Ward’s method produced the highest distance measure between x_i and x_j.

Innovation activities in the service sector across EU states: Similarities or differences?

Until 2005, service sector innovation had been a far less popular area of scholarly interest than studies on the industrial sector, the publication in 2005 of the methodological guidelines known as the Oslo Manual, in which it was noted that in addition to technological advancements in the economy, non-technological innovations might also be pursued in economic practice. Since then, although the popularity and dynamics of dedicated research have successively increased, this field of inquiry has not yet been sufficiently explored. Therefore, the aim of this study was to identify and assess the degree of homogeneity among EU countries in terms of innovation activity in the service sector. The hypothesis adopted for this study stated that there are countries within the EU that can be divided into clusters with respect to service sector innovation activity. Bearing in mind the objective and seeking to verify the hypothesis, this research used a critical review of literature, indicator-oriented analysis as well as quantitative statistical methods, among others, while relying on the most up-to-date data from the Eurostat database on innovation activity (as of 5.07.2022). The objective was accomplished using fifteen indicators which describe the effects of innovation activity, which were classified into four groups: innovation activity, types of implemented innovations, types of product innovation, and types of business process innovation. The research proceeded in three stages: identification of groups (clusters) in the area of the surveyed European countries which demonstrated the homogeneity of analysed indicators using Ward’s agglomerative clustering method; identification of significant differences between the defined groups for each indicator carried out, using Student's t-test; an assessment of the degree of homogeneity in the scope of the surveyed EU countries with respect to innovation activity in the service sector by means of the methods outlined above. On the basis of the conducted investigations and the adopted methodology, it may be concluded, first, that the service sectors of the surveyed EU countries can be divided into three homogeneous groups in terms of innovation activity. Second, the similarity within each group was confirmed by Levene’s test, the results of which demonstrated homogeneity of variance for thirteen out of the fifteen analysed indicators. Thirdly, statistical analysis of the findings made it possible to identify significant differences between the defined groups of countries, with the exception of country clusters I and II with respect to three product innovation indicators, as well as country groups II and III relative to three business process innovation indicators, where no significant differences occurred. The conclusions from the study warrant confirmation of the baseline hypothesis, which assumed that the surveyed EU countries could be divided into clusters in terms of service sector innovation activity. The research and the analyses indicated that a homogeneous division of the surveyed EU countries into individual groups in terms of innovative activities in the service sector might offer a point of departure for further, in-depth analysis of the issues addressed in the study. Some of the compelling directions for future investigations include the determination of homogeneity and differences across the industries of the sector in question, as well as the creation of a ranking of countries in line with the analysed indicators. Approaching the issue in this manner will enable a detailed assessment of the innovation activity in the service sector and its internal structures, as well as provide grounds for validating the cluster division presented in this paper.

Assessment of genetic diversity in Indian sesame (Sesamum indicum L.) germplasm

Principal component analysis and hierarchical clustering were carried out for 151 Indian sesame germplasm for five quantitative traits, viz., days to fifty percent flowering, days to maturity, plant height (cm), number of capsules per plant and seed yield per plant (g). Among five principal components, PC I and PC II alone recorded above 1 eigenvalue. The first two principal components accounted for 62.45% of cumulative total variations. PC I contributed the maximum variability of 40%, followed by PC II (21.68%) of the total variation. Days to fifty percent flowering and days to maturity had high and negative loading values in PC I. In the case of PC II, plant height had the highest and negative loading value. Among characters, days to flowering and maturity had a positive association with seed yield per plant. Similarly, plant height had a positive association with seed yield per plant. Hence these traits may be considered as selection indices for the improvement of seed yield. Sesame germplasm accessions were grouped into eight distinct clusters using the agglomerative clustering method. The hierarchical cluster analysis indicates that cluster I had the highest genotypic members (58) followed by cluster II (50 genotypes). Cluster IV holds 15 genotypes and Cluster V has 11 genotypes. Clusters III and VI had seven genotypes for each cluster. However, cluster VII holds only one genotype and cluster VIII has two genotypes. In the present study, the level of variability in sesame germplasm was found to be similar in both principal component analysis and hierarchical cluster method.

Improving crop rotation classification using a random forest model incorporating spatial heterogeneity

Accurate and timely classification of crop rotations is essential to confront the issues of agricultural management and food crisis. Crop growth conditions generally exhibit a strong spatial heterogeneity pattern, resulting in crop growth characteristics varying with locations, limiting the classification accuracy of crop rotation. To overcome this limitation, an improved method named random forest based on rotation zoning strategy (RF_RZS) that classifies crop rotations under the consideration of spatial heterogeneity is proposed. In RF_RZS, the regionalization with dynamically constrained agglomerative clustering and partitioning method is used to adaptively mine the spatial homogeneous subzones of soil organic carbon (SOC), which represent the spatial heterogeneity pattern of crop rotation given the highly correlated relationship between crop rotation and SOC. The Boruta algorithm is employed to select the optimal feature subset within each subzone. A random forest classification method is applied to categorize crop rotations within the subzones. Two integrated indexes, OA_OZ and Kc_OZ, are proposed to evaluate the comprehensive performance of RF_RZS. Furthermore, a series of traditional methods is employed for comparison and evaluation. Results demonstrate that the OA_OZ and Kc_OZ of the RF_RZS method are 0.90 and 0.88, respectively, which are increased by 3–23% and 3–27%, respectively, compared with those of traditional classification methods. This research involves mixed cropping pattern of multiple rotational crops, and the proposed methodology can provide an effective guide for the classification of rotational crops with a fragmented distribution and a complex cropping structure.

Inferring “high-frequent” mixed urban functions from telecom traffic

Precise distinction of mixed functions on urban land is essential for urban studies and planning, while existing methods are limited by high sampling bias, low observation frequency, and lack of semantic information in common data sources. In this paper, we introduce a new proxy for human behavior, the telecom traffic data as a remedy to the above limitations, and present an analytical framework which utilizes anonymized and aggregated telecom traffic data to infer mixed urban functions at spatiotemporal granularities as fine as buildings and hours. A time-series decomposition method is designed to map the mixture of urban functions, which is further refined by a hierarchical agglomerative clustering method taking urban textures as an additional source of information. In a case study in Shenzhen, China, we find the function of urban buildings can be decomposed into the mixture of three basic functions, namely dwelling, work, and recreation. We further find that the introduction of urban texture information helps identify particular forms of functional combination, which indicate special-function buildings such as urban villages and roadside shops. This study implies ways to improve urban management through methodological contributions in mixed urban function identification alongside the introduction of the telecom traffic, a kind of “high-frequency” urban data, and also helps inspire a rethinking of the form/function dichotomy in the era of “High-frequent” cities.

Identifying Locations for Transcranial Magnetic Stimulation Coil Placement For Deep Brain Structures

Stimulation of deep brain structures with transcranial magnetic stimulation (TMS) is the premier mode of non-invasive brain stimulation (NIBS). Currently, there is little support for the application of TMS in modulating deeper brain structures. This study validates the use of TMS for a set of 8 deep brain structures shown to alleviate symptoms of certain conditions when stimulated. 200 sets of T1-weighted and T2-weighted MRIs were processed using the SimNIBS headreco pipeline. Then, 200 TMS optimization simulations were run for each brain structure, which generated 200 3D coordinates for each structure representing the best TMS coil placement locations on the scalp for stimulation. This data was analyzed using K-means and agglomerative clustering methods to find the overall best coil placement locations. Overall, this study validates the use of TMS to stimulate these brain structures and finds the optimal parameters for stimulation.

Proposal of a Visualization System for a Hierarchical Clustering Algorithm: The Visualize Proximity Matrix

Visual data mining is a new and effective strategy for dealing with the growing phenomenon of information overload. There is an urgent need for effective visual data mining because the growth of data sets from different domains and sources has made exploring, managing and analyzing vast volumes of data increasingly difficult. Identifying location-based trends and anomalies in data from a numeric output is challenging. The outputs of data mining procedures are often quite difficult to interpret. There is no human input to data exploration or direct involvement in the data mining process. Thus, it is difficult to analyze, explore, understand and view the data. In data mining, massive data sets are clustered using an iterative process that includes human input but data mining algorithms can miss some knowledge and observations. In addition, many users cannot take the appropriate action to address problems or opportunities because the huge amounts of data prevent them from staying aware of what is happening in and around their environments. Visual data mining can help in dealing with information overload and it is effective in analyzing large complex data sets. Visual data mining assists users in gaining a deep visual understanding of data. In the information age, users need to study and observe vast amounts of data to acquire important knowledge, and thus the need for visual and interactive analytical tools is particularly pressing. Visual cluster analysis has long utilized shaded similarity matrices, and this study investigated how they can be used in clustering visualization. We focused on proposing an agglomerative hierarchical clustering method. Ensemble cluster visualization is also presented for handling large data sets. This study proposes the adoption of a shaded similarity matrix to visually cluster knowledge discovered using data mining. Using the technology acceptance model as the measuring tool, we questioned respondents to evaluate the visualization prototype. The findings demonstrated that the visualization was effective and easy to use, and satisfied users.

Insights in Hierarchical Clustering of Variables for Compositional Data

R-mode hierarchical clustering is a method for forming hierarchical groups of mutually exclusive subsets of variables. This R-mode cluster method identifies interrelationships between variables which are useful for variable selection and dimension reduction. Importantly, the method is based on metric elements defined on the sample space of variables. Consequently, hierarchical clustering of compositional parts should respect the particular geometry of the simplex. In this work, the connections between concepts such as distance, cluster representative, compositional biplot, and log-ratio basis are explored within the framework of the most popular R-mode agglomerative hierarchical clustering methods. The approach is illustrated in a paleoecological study to identify groups of species sharing similar behavior.

Identification of the energy crisis in the EU electricity markets

In the second half of 2021 - the first half of 2023, the EU electricity markets were under unprecedented pressure caused by soaring gas prices and aggravated by other fundamental factors. This period corresponds to the energy crisis in the EU electric power sector. However, all EU electricity markets responded to the energy crisis differently. This paper proposes a methodology for revealing the energy crisis in the electricity market, which is based on the tools of descriptive statistics, explanatory data analysis and k-means clustering which allows to identification duration and phases of the energy crisis such as escalation, peak, and recovery phases. Using this methodology, the energy crisis was identified in the EU electricity markets (by separate bidding zones), and as a result, the starting and ending points of the energy crisis, as well as the changing by phases, were determined. Additionally, the use of hierarchical and agglomerative clustering methods made it possible to divide all EU electricity markets into 9 clusters by the evolution of the energy crisis and identify unaffected, resilient vulnerable markets to the energy crisis, and markets with their own dynamics.

Prediction of FKRTL Services Diagnosed with Type 2 Diabetes Mellitus Using the Hierarchical Agglomerative Clustering Time Series Method

Diabetes Mellitus (DM) is a disease characterized by hyperglycemia or increased blood sugar levels and metabolic disorders. DM is a disease with a high prevalence rate in Indonesia. DM is included in the 10 most frequent Advanced Referral Health Facility (FKRTL) visits diagnoses. This research aims to find out provincial groupings based on data from FKRTL visits of participants diagnosed with T2DM using the Hierarchical Agglomerative Clustering method (single linkage, complete linkage, average linkage) and to find out predictions of participants diagnosed with T2DM at FKRTL visits using the Autoregressive Integrated Moving Average (ARIMA) method. The method used is Hierarchical Agglomerative Clustering Time Series. This research shows that the best algorithm is average linkage with a number of clusters of two. Forecasting provides a pattern that tends to decline.

The Impact of Long-Term Conditions and Comorbidity Patterns on COVID-19 Infection and Hospitalisation: A Cohort Study

Introduction: Older adults are more vulnerable to COVID-19 infections; however, little is known about which comorbidity patterns are related to a higher risk of COVID-19 infection. This study investigated the role of long-term conditions or comorbidity patterns on COVID-19 infection and related hospitalisations. Methods: This study included 4,428 individuals from Waves 8 (2016−2017) and 9 (2018−2019) of the English Longitudinal Study of Ageing (ELSA) who also participated in the ELSA COVID-19 Substudy in 2020. Comorbidity patterns were identified using an agglomerative hierarchical clustering method. The relationships between comorbidity patterns or long-term conditions and COVID-19-related outcomes were examined using multivariable logistic regression. Results: Among a representative sample of community-dwelling older adults in England, those with cardiovascular disease (CVD) and complex comorbidities had an almost double risk of COVID-19 infection (OR = 1.87, 95% CI = 1.42−2.46) but not of COVID-19-related hospitalisation. A similar pattern was observed for the heterogeneous comorbidities cluster (OR = 1.56, 95% CI = 1.24−1.96). The individual investigations of long-term conditions with COVID-19 infection highlighted primary associations with CVD (OR = 1.46, 95% CI = 1.23−1.74), lung diseases (OR = 1.40, 95% CI = 1.17−1.69), psychiatric conditions (OR = 1.40, 95% CI = 1.16−1.68), retinopathy/eye diseases (OR = 1.39, 95% CI = 1.18−1.64), and arthritis (OR = 1.27, 95% CI = 1.09−1.48). In contrast, metabolic disorders and diagnosed diabetes were not associated with any COVID-19 outcomes. Conclusion: This study provides novel insights into the comorbidity patterns that are more vulnerable to COVID-19 infections and hospitalisations, highlighting the vulnerability of those with CVD and other complex comorbidities. These findings facilitate crucial new evidence that should be considered for appropriate screening measures and tailored interventions for older adults in the ongoing global outbreak.

Antineutrophil cytoplasmic antibody-associated vasculitis classification by cluster analysis based on clinical phenotypes: a single-center retrospective cohort study.

Antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) refers to a group of small vessel inflammatory disorders. Overlapping clinical phenotypes of AAV subgroups continually provoke controversies over their diagnostic and classification criteria. Using the agglomerative hierarchical clustering method, we classified 210 Korean patients diagnosed with AAV into mutually exclusive clusters according to Birmingham Vasculitis Activity Score items, ANCA specificity, sex, and age. We analyzed the resulting clusters' outcomes to investigate the clinical significance of the classification. We proposed a distance-based algorithm of patient assignment and explored its clinically relevant modification. In total, 116 patients (55%) had microscopic polyangiitis, 53 (25%) had granulomatosis with polyangiitis, and 42 (20%) had eosinophilic granulomatosis with polyangiitis. Our model grouped the patients into five clusters, namely, "limited proteinase 3 (PR3)-ANCA vasculitis," "generalized PR3-ANCA vasculitis," "ANCA-negative vasculitis," "renal-limited vasculitis," and "myeloperoxidase-ANCA vasculitis." Patients clustered under "generalized PR3-ANCA vasculitis" had a higher relapse rate (hazard ratio [HR] = 2.12, P = 0.067). The incidence of end-stage renal disease was higher in patients belonging to the "renal-limited vasculitis" cluster (HR=1.50, P=0.03), and those in the "ANCA-negative vasculitis" cluster experienced a relatively milder clinical course of AAV (mortality = 0). Because the clusters were naturally derived from their distinguished phenotypes and have different clinical courses, our clustering method may be a more clinically relevant classification system for AAV, revealing its phenotypic diversity. We also proposed a simple and intuitive distance-based assignment algorithm, which can be easily modified according to specific clinical needs. Key Points • In this study with a single-center AAV cohort, we showed that AAV can be divided into five distinct subclasses with different disease courses based on the clinical and laboratory features of the patients. • Our study revealed ethnic differences in AAV manifestation and suggests that physicians may need to analyze their own AAV patients to assess the disease status of AAV patients. • We proposed a distance-based cluster membership assignment method that can be clinically modified to fit the specific purpose of grouping patients.

Connectomic insight into unique stroke patient recovery after rTMS treatment.

An improved understanding of the neuroplastic potential of the brain has allowed advancements in neuromodulatory treatments for acute stroke patients. However, there remains a poor understanding of individual differences in treatment-induced recovery. Individualized information on connectivity disturbances may help predict differences in treatment response and recovery phenotypes. We studied the medical data of 22 ischemic stroke patients who received MRI scans and started repetitive transcranial magnetic stimulation (rTMS) treatment on the same day. The functional and motor outcomes were assessed at admission day, 1 day after treatment, 30 days after treatment, and 90 days after treatment using four validated standardized stroke outcome scales. Each patient underwent detailed baseline connectivity analyses to identify structural and functional connectivity disturbances. An unsupervised machine learning (ML) agglomerative hierarchical clustering method was utilized to group patients according to outcomes at four-time points to identify individual phenotypes in recovery trajectory. Differences in connectivity features were examined between individual clusters. Patients were a median age of 64, 50% female, and had a median hospital length of stay of 9.5 days. A significant improvement between all time points was demonstrated post treatment in three of four validated stroke scales utilized. ML-based analyses identified distinct clusters representing unique patient trajectories for each scale. Quantitative differences were found to exist in structural and functional connectivity analyses of the motor network and subcortical structures between individual clusters which could explain these unique trajectories on the Barthel Index (BI) scale but not on other stroke scales. This study demonstrates for the first time the feasibility of using individualized connectivity analyses in differentiating unique phenotypes in rTMS treatment responses and recovery. This personalized connectomic approach may be utilized in the future to better understand patient recovery trajectories with neuromodulatory treatment.