Mugen Aerospace

Mugen Aerospace Contact information, map and directions, contact form, opening hours, services, ratings, photos, videos and announcements from Mugen Aerospace, Aerospace Company, 35 Calgary Street, Glasgow.

Time Series Analysis (TSA) is a branch of statistics that deals with analyzing data points collected over time. TSA focu...
12/12/2023

Time Series Analysis (TSA) is a branch of statistics that deals with analyzing data points collected over time. TSA focuses on identifying trends, patterns, and seasonality in the data. The output of this process can then be used to make predictions about future values, understand relationships between variables, and gain insights into the underlying processes that generate the data.

Some key features of this analysis are the following:

1. Regular intervals
2. Data order
3. Complex data point relationships

Using MODIS bands, sur_refl_b04 (green) & sur_refl_b06 (short-wave infrared), we determine the Normalized Difference Snow Index (NDSI) that calculates the snow presence, described in a single pixel, in the country of Poland and its voivodeships.

We first plot the univariate time series values for the whole country from 2019 to 2022 by aggregating the computed values in arrays, and converting them to data frames. Then, we use two nested functions for nested mapping of univariate time series calculations for each voivodeship, and extract them in vectors to graph the data frames.

In finding similar trends between the voivodeships, we use K-means clustering method from scikit-learn and tslearn libraries built specifically for time series computations.

To synthesize, TSA is often used for forecasting future values based on historical data. Forecasting helps in making predictions about future trends and patterns of an event to assimilate.

—
Message us for the full source code/prediction model 🛰️

Unsupervised machine learning algorithms are schemes that learn from unlabelled data. These are often used for explorato...
21/11/2023

Unsupervised machine learning algorithms are schemes that learn from unlabelled data. These are often used for exploratory data analysis, customer segmentation, and anomaly detection.

One type of this algorithm is K-means clustering which partitions a set of data points into a predefined number of clusters (k). It aims to group similar data points together by minimizing the within-cluster variance, which is the average squared distance between data points within a cluster. This distance-based algorithm works iteratively, starting with randomly assigned cluster centroids and then reassigning data points to the nearest centroid until convergence is reached.

Here, we tested K-means on the city of New York where there is a large number of neighborhoods and a diverse population. Three predefined numbers of clusters were applied to the city boundary which were also visualized to identify the patterns that formed as k increased.

Overall, K-means clustering is a powerful tool for data exploration and analysis. It is, however, important to be aware of its limitations and to use it with caution. When applying K-means clustering, it is necessary to experiment with different values of k and different initialization methods, and to be careful when interpreting the results.

---
Message us for the full source code/prediction model 🛰️

Support Vector Machine (SVM) is a type of supervised machine learning algorithm that can be used for both classification...
13/11/2023

Support Vector Machine (SVM) is a type of supervised machine learning algorithm that can be used for both classification and regression problems. SVM works by finding a hyperplane in a high-dimensional space using a kernel function that separates the data points into two classes. The hyperplane is chosen such that it maximizes the margin between the two classes.

While SVM is vigorous to outliers and noise in the data, however, it can be computationally expensive to train, especially for large satellite datasets.

In this example, we see how SVM is applied to predict the present biomass carbon density (BCS) in Sichuan Province, China, using a 2010 satellite dataset. BCS is an important metric for understanding the role of biomass in the carbon cycle. It is also used to assess the potential of biomass for carbon sequestration and bioenergy production.

Multiple bands of Landsat 9 imagery were utilized for model prediction, and the terrestrial carbon storage band of 2010 global BCS by World Conversation Monitoring Centre (WCMC) was used for model training.

The Root Mean Squared Error (RMSE) obtained was 0.0935 for training data and 0.1666 for validation data which indicated predictions that are closer to ground truth.

--
Message us for the full source code/prediction model 🛰️

Random Forest (RF) is an ensemble learning algorithm that combines the predictions of multiple decision trees to produce...
07/11/2023

Random Forest (RF) is an ensemble learning algorithm that combines the predictions of multiple decision trees to produce a more accurate prediction. RF works by building a forest of decision trees, where each tree trains from a random subset of the data and uses a random subset of features.

RF algorithm has several advantages over other machine learning algorithms, including:

1. Accuracy: RFs are typically very accurate, even on complex datasets.

2. Robustness: RFs are robust to overfitting and noise in the data.

3. Interpretability: RFs are relatively interpretable, which means that it is possible to understand how the model is making its predictions.

4. Scalability: RFs can be scaled to train on very large datasets.

Providing a Landsat 9 imagery within a circular boundary inside Melbourne, Australia, an RF ensemble model was labelled using MODIS dataset (Land_Cover_Type_1 band) and classified applying the International Geosphere-Biosphere Programme (IGBP) palette.

The prediction achieved an overall accuracy of 94.6%

--
Message us for the full source code/prediction model 🛰️

Machine Learning is a powerful tool with a wide range of applications and benefits when applied to satellite data. One o...
04/11/2023

Machine Learning is a powerful tool with a wide range of applications and benefits when applied to satellite data. One of its algorithms is the CART (Classification and Regression Tree) predictive model that categorizes data points (i.e., training samples) within a geometric boundary into predefined categories for data visualization.

Here is a sample of a CART classifier that predicts in three random geographical areas within the Philippines.

Training points are provided by Google Earth Engine for urban, vegetation, and water classes. The images are captured using Landsat 9 (30m resolution) and overlaid with a classification layer encapsulating the following information:

Class 0 for Urban represented by red
Class 1 for Vegetation represented by green
Class 2 for Water represented by blue

Analyzing vast and complex datasets from satellites has the potential to revolutionize how we respond to agricultural enhancement, disaster management, and infrastructure planning. It enables more timely, accurate, and data-driven decision-making.

—
Message us for the full source code/prediction model 🛰️

Address

35 Calgary Street
Glasgow
G40XG

Website

Alerts

Be the first to know and let us send you an email when Mugen Aerospace posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to Mugen Aerospace:

Shortcuts

Share

Category