Glm

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:41Z ----------------------------------------------------------------

You have 2 design matrix in one so the plot with time at the bottom is a bit confusing as it reference both filter from 0 to 10? Also axis legend is missing units (sec?)

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:42Z ----------------------------------------------------------------

Nice! You should make a comment and note how the R2 is higher for black flashes here. This is already a result

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

What is going on here. Maybe explain this cell a bit.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

I will start with an introduction here :

Model fitting is useful if you are expecting neuronal activity to follow a certain behavior. In the case of our Poisson model, we impose a certain response formula and fit the data to it. It can be sometimes useful to compare this fit with a more open-ended selection of cells: for instance, selecting cells simply based on whether the cell fired above a certain threshold. The combination of both approaches can allow to select cells to which fitting a model yield meaningful insights. In some cases, a non-responsive cell could provide model parameters that are hard to interpret. So it is always judicious to combine approaches.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:44Z ----------------------------------------------------------------

Add some docstring to explain this function

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:45Z ----------------------------------------------------------------

Explain your results a little more. Those are cells selected for responses.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:42Z ----------------------------------------------------------------

Nice! You should make a comment and note how the R2 is higher for black flashes here. This is already a result

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

What is going on here. Maybe explain this cell a bit.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

I will start with an introduction here :

Model fitting is useful if you are expecting neuronal activity to follow a certain behavior. In the case of our Poisson model, we impose a certain response formula and fit the data to it. It can be sometimes useful to compare this fit with a more open-ended selection of cells: for instance, selecting cells simply based on whether the cell fired above a certain threshold. The combination of both approaches can allow to select cells to which fitting a model yield meaningful insights. In some cases, a non-responsive cell could provide model parameters that are hard to interpret. So it is always judicious to combine approaches.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:44Z ----------------------------------------------------------------

Add some docstring to explain this function

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:45Z ----------------------------------------------------------------

Explain your results a little more. Those are cells selected for responses.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

What is going on here. Maybe explain this cell a bit.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

I will start with an introduction here :

Model fitting is useful if you are expecting neuronal activity to follow a certain behavior. In the case of our Poisson model, we impose a certain response formula and fit the data to it. It can be sometimes useful to compare this fit with a more open-ended selection of cells: for instance, selecting cells simply based on whether the cell fired above a certain threshold. The combination of both approaches can allow to select cells to which fitting a model yield meaningful insights. In some cases, a non-responsive cell could provide model parameters that are hard to interpret. So it is always judicious to combine approaches.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:44Z ----------------------------------------------------------------

Add some docstring to explain this function

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:45Z ----------------------------------------------------------------

Explain your results a little more. Those are cells selected for responses.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:43Z ----------------------------------------------------------------

I will start with an introduction here :

Model fitting is useful if you are expecting neuronal activity to follow a certain behavior. In the case of our Poisson model, we impose a certain response formula and fit the data to it. It can be sometimes useful to compare this fit with a more open-ended selection of cells: for instance, selecting cells simply based on whether the cell fired above a certain threshold. The combination of both approaches can allow to select cells to which fitting a model yield meaningful insights. In some cases, a non-responsive cell could provide model parameters that are hard to interpret. So it is always judicious to combine approaches.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:44Z ----------------------------------------------------------------

Add some docstring to explain this function

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:45Z ----------------------------------------------------------------

Explain your results a little more. Those are cells selected for responses.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:44Z ----------------------------------------------------------------

Add some docstring to explain this function

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:45Z ----------------------------------------------------------------

Explain your results a little more. Those are cells selected for responses.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:45Z ----------------------------------------------------------------

Explain your results a little more. Those are cells selected for responses.

review-notebook-app[bot] commented 3 months ago

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.

[review-notebook-app[bot]]

View / edit / reply to this conversation on ReviewNB

jeromelecoq commented on 2024-04-26T00:23:46Z ----------------------------------------------------------------

You need to highlight this result a little better.

The first thing I notice is that filter can be positive or negative. You should explain that the cell firing in the poisson model is a combination of basal firing rate and of the cummulative application of the filter on the past.

SO

A strong negative value at the onset of the filter means a recent stimuli will INHIBIT firing. And vice versa.

Comparing the average response with the filters: You see that the cell that fire late tend to have negative values at the onset of the stimulus and strong positive values later, and vice versa.

This little analysis allow to read both columns and "interpret" your results.

You should also say that here you selected the cells with strong responses to which the filter fit should yield more meaningful filters, ie. we are not fitting the noise.

• © Githubissues.
• Githubissues is a development platform for aggregating issues.