Month: November 2023

Hii,

In todays  analysis, that I learned about the time series models for predecting the future data.

1. SARIMA (Seasonal Autoregressive Integrated Moving Average): This model extends the ARIMA model by incorporating seasonality. SARIMA is useful in datasets where seasonal patterns are prominent, and it includes additional seasonal parameters to account for these patterns.
2. VAR (Vector Autoregression): The VAR model is used for multivariate time series data, where the system captures the linear interdependencies among multiple variables. It’s particularly useful when you want to model and predict systems where variables influence each other.
3. LSTM (Long Short-Term Memory): LSTM is a type of recurrent neural network (RNN) particularly effective in learning order dependence in sequence prediction problems. This model is well-suited for time series data where there are long-term dependencies or patterns.
4. ARIMA (Autoregressive Integrated Moving Average): ARIMA is one of the classic models used for time series forecasting. It combines autoregressive features with moving averages and integrates differencing to make the time series stationary, making it effective for a wide range of datasets.

Today I analysis the economic indicators from a dataset of anylize Boston. There are 264 datasets. This is a legacy dataset of economic indicators tracked monthly between January 2013 and December 2019 from the Boston Planning and Development Authority (BPDA) tasked with planning for and guiding inclusive growth within the City of Boston. And I studied the example of Time series data.

In todays class we discussed about the time series data and forecasting. The  time series data refers to a sequence of data points collected or recorded at regular time intervals. This kind of data is prevalent in various fields such as economics, finance, environmental science, and more. The key characteristics of time series data include trends, seasonality, and cyclic patterns.

Sequential Nature: The primary characteristic of time series data is its sequential order. Unlike other types of data, the order in which the data points are arranged is crucial because it reflects the progression of time.

Components of Time Series:

• Trend: It represents the long-term progression of the series. Trends can be upward, downward, or even sideways over time.
• Seasonality: These are patterns that repeat at regular intervals, such as hourly, daily, weekly, monthly, or annually. For example, increased ice cream sales during summer months.
• Cyclic Patterns: Unlike seasonal patterns, cyclic patterns occur over longer periods and are not fixed to a particular time frame. They are often related to business or economic cycles.
• Irregular or Random Components: These are unforeseeable variations that are not part of the trend, seasonality, or cyclic components. They often result from unforeseen or random events.

MTH522_2

In today’s lesson, I learned about decision trees, which are visual models of the decision-making process. Imagine them as a flow of queries and choices culminating in a definitive result. The process begins with an initial query, and with each subsequent answer, you proceed along the pathways until you reach the final outcome.

• Root Node: This is the first node of the tree where the data splits. It represents the entire dataset, which then gets divided into two or more homogeneous sets.
• Splitting: It is the process of dividing a node into two or more sub-nodes based on certain conditions.
• Decision Node: When a sub-node splits into further sub-nodes, it’s called a decision node.
• Leaf/Terminal Node: Nodes that do not split any further are called leaves or terminal nodes. They represent the output or the decision taken after computing all attributes.
• Branches or Sub-Trees: A section of the entire tree is called a branch or sub-tree.
• Parent and Child Node: A node, which is divided into sub-nodes, is called the parent node of the sub-nodes, whereas the sub-nodes are the children of the parent node.
•   And we done the logistic regression to given data and we written our code.

Today we analysis the data of washington post data and we observe  there’s a positive correlation between age and the not flee status. This might indicate that older individuals are less likely to flee or resist, but without a specific correlation coefficient, it’s hard to quantify the strength of this relationship. There is a negative correlation between the not flee status and incidents related to mental illness. Even though the correlation is weak, this could suggest that individuals with mental illnesses might be slightly more likely to flee. However, this connection is not robust.

He strength of these correlations, as you mentioned, is weak. This means that while there might be some relationships between flee status and other factors, these relationships are not dominant or necessarily predictive. It would be essential to consider other potential confounding variables or factors that could impact these relationships. Additionally, correlation does not imply causation. So, even if there’s a relationship between two variables, one doesn’t necessarily cause the other.