Linear trendline:wrong equation

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with the
right equation???

Thanks

Jan M.

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that your
x-axis was categorical instead of numeric, and that what you provided for the
x-axis was a set of category labels that may or may not have numeric values.
Why Excel would offer to fit a trendline in that circumstance is a mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with the
right equation???

Thanks

Jan M.

Similarly, the bar and column charts assume that the x-axis is non-numerica
categories to be treated as 1,2,3,... if a trendline is requested.

Excel provides no chart type that plots bar heights/lengths against a
numberic axis.

Jerry

"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that your
x-axis was categorical instead of numeric, and that what you provided for the
x-axis was a set of category labels that may or may not have numeric values.
Why Excel would offer to fit a trendline in that circumstance is a mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with the
right equation???

Thanks

Jan M.

Jerry,

thanks for your reply!

You are right about excel using 1,2,3 as x-values: I hadn't figured that one
out!

What do you mean by an "XY (Scatter)" chart? I'm still missing something
he What should I do differently to have excel give me the right equation?

Thanks

"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that your
x-axis was categorical instead of numeric, and that what you provided for the
x-axis was a set of category labels that may or may not have numeric values.
Why Excel would offer to fit a trendline in that circumstance is a mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with the
right equation???

Thanks

Jan M.

Jerry,

I think I've figured out what your meant by "xy scatter"!
I modified my data table the following way and got the right results:

X Y
73 6.6
74
75
76
77
78 5.8
.... ...

Thanks for your help.

Jan M.



"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that your
x-axis was categorical instead of numeric, and that what you provided for the
x-axis was a set of category labels that may or may not have numeric values.
Why Excel would offer to fit a trendline in that circumstance is a mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with the
right equation???

Thanks

Jan M.

What Jerry means is don't use a column chart. It is the wrong kind of chart
to use to show relationships between two variables, even if you are not
trying to generate statistics on it. Rebuild the chart and select one of the
XY subtypes in step 1 of the chart wizard, or convert the chart using Chart
Type on the Chart menu. There's no need to mess around with the data range.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______

"Jan M." wrote in message
...
Jerry,

I think I've figured out what your meant by "xy scatter"!
I modified my data table the following way and got the right results:

X Y
73 6.6
74
75
76
77
78 5.8
... ...

Thanks for your help.

Jan M.



"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that
your
x-axis was categorical instead of numeric, and that what you provided for
the
x-axis was a set of category labels that may or may not have numeric
values.
Why Excel would offer to fit a trendline in that circumstance is a
mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly
calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which
seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the
following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a
rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with
the
right equation???

Thanks

Jan M.

Jon,

thanks for your input.

I've tried changing the chart type with no results. What is it that you call
an XY subtype? It may seems trivial to you, but I assure you that it is not
to me (maybe because I'm using a french version).

Thanks.

Jan M.


"Jon Peltier" wrote:

What Jerry means is don't use a column chart. It is the wrong kind of chart
to use to show relationships between two variables, even if you are not
trying to generate statistics on it. Rebuild the chart and select one of the
XY subtypes in step 1 of the chart wizard, or convert the chart using Chart
Type on the Chart menu. There's no need to mess around with the data range.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______

"Jan M." wrote in message
...
Jerry,

I think I've figured out what your meant by "xy scatter"!
I modified my data table the following way and got the right results:

X Y
73 6.6
74
75
76
77
78 5.8
... ...

Thanks for your help.

Jan M.



"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that
your
x-axis was categorical instead of numeric, and that what you provided for
the
x-axis was a set of category labels that may or may not have numeric
values.
Why Excel would offer to fit a trendline in that circumstance is a
mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly
calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which
seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the
following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a
rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with
the
right equation???

Thanks

Jan M.

Okay, I'll try to type more slowly. That's a joke.

In step 1 of the Chart Wizard, or in the Chart Type dialog, the top five
chart types in the left list are Column, Bar, Line, Pie, and XY. Select the
fifth (XY) as the chart type. Even in the French version, I'm sure the icons
are in the same order.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______

"Jan M." wrote in message
...
Jon,

thanks for your input.

I've tried changing the chart type with no results. What is it that you
call
an XY subtype? It may seems trivial to you, but I assure you that it is
not
to me (maybe because I'm using a french version).

Thanks.

Jan M.


"Jon Peltier" wrote:

What Jerry means is don't use a column chart. It is the wrong kind of
chart
to use to show relationships between two variables, even if you are not
trying to generate statistics on it. Rebuild the chart and select one of
the
XY subtypes in step 1 of the chart wizard, or convert the chart using
Chart
Type on the Chart menu. There's no need to mess around with the data
range.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______

"Jan M." wrote in message
...
Jerry,

I think I've figured out what your meant by "xy scatter"!
I modified my data table the following way and got the right results:

X Y
73 6.6
74
75
76
77
78 5.8
... ...

Thanks for your help.

Jan M.



"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that
your
x-axis was categorical instead of numeric, and that what you provided
for
the
x-axis was a set of category labels that may or may not have numeric
values.
Why Excel would offer to fit a trendline in that circumstance is a
mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly
calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which
seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the
following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a
rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up
with
the
right equation???

Thanks

Jan M.

Jon and Jerry,

I've found the XY (Scatter)! In a french version it is called "Nuages de
points" (Clouds of dots). Very poetic isn't it?

Thanks to both of you.

Jan M.



"Jon Peltier" wrote:

What Jerry means is don't use a column chart. It is the wrong kind of chart
to use to show relationships between two variables, even if you are not
trying to generate statistics on it. Rebuild the chart and select one of the
XY subtypes in step 1 of the chart wizard, or convert the chart using Chart
Type on the Chart menu. There's no need to mess around with the data range.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______

"Jan M." wrote in message
...
Jerry,

I think I've figured out what your meant by "xy scatter"!
I modified my data table the following way and got the right results:

X Y
73 6.6
74
75
76
77
78 5.8
... ...

Thanks for your help.

Jan M.



"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that
your
x-axis was categorical instead of numeric, and that what you provided for
the
x-axis was a set of category labels that may or may not have numeric
values.
Why Excel would offer to fit a trendline in that circumstance is a
mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly
calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which
seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the
following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a
rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up with
the
right equation???

Thanks

Jan M.

Very nice. Is that what Frenchmen call it, or is that just Microsoft's
unique translation?

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______


"Jan M." wrote in message
...
Jon and Jerry,

I've found the XY (Scatter)! In a french version it is called "Nuages de
points" (Clouds of dots). Very poetic isn't it?

Thanks to both of you.

Jan M.



"Jon Peltier" wrote:

What Jerry means is don't use a column chart. It is the wrong kind of
chart
to use to show relationships between two variables, even if you are not
trying to generate statistics on it. Rebuild the chart and select one of
the
XY subtypes in step 1 of the chart wizard, or convert the chart using
Chart
Type on the Chart menu. There's no need to mess around with the data
range.

- Jon
-------
Jon Peltier, Microsoft Excel MVP
Peltier Technical Services
Tutorials and Custom Solutions
http://PeltierTech.com/
_______

"Jan M." wrote in message
...
Jerry,

I think I've figured out what your meant by "xy scatter"!
I modified my data table the following way and got the right results:

X Y
73 6.6
74
75
76
77
78 5.8
... ...

Thanks for your help.

Jan M.



"Jerry W. Lewis" wrote:

Use an "XY (Scatter)" chart.

When you selected a "Line" chart, you (by definition) told Excel that
your
x-axis was categorical instead of numeric, and that what you provided
for
the
x-axis was a set of category labels that may or may not have numeric
values.
Why Excel would offer to fit a trendline in that circumstance is a
mystery to
me, but when it does, it uses x-values of 1,2,3,... and correctly
calculates
the regression of y against those assumed x-values.

Jerry

"Jan M." wrote:

Hi,

I created a bar chart in Excel from the following data:

X Y
73 6.6
78 5.7
86 4.8

The SLOPE and the INTERCEPT functions returned -0.136 and 16.4448
respectively. The resulting equation is Y = -0.136X + 16.448 which
seems good
enough to me.

Then I added a linear trendline to the chart. Excel displayed the
following
equation: Y = -0.9X + 7.5, R ^2 =1 which is way off (and it's not a
rounding
problem)!!!

The data seemed farly linear to me, how come Excel can't come up
with
the
right equation???

Thanks

Jan M.